Development and processing of solid oxide fuel cell materialsetheses.bham.ac.uk/5861/1/Pike15PhD.pdf · processing of novel materials for use within solid oxide fuel cells. A range

i

Development and Processing of Solid

Oxide Fuel Cell Materials

By

Thomas William Pike

i. Title

Primary Supervisor: Professor Peter R. Slater

Secondary Supervisor: Professor Robert Steinberger-Wilckens

A thesis submitted to The University of Birmingham for the degree of

Doctor of Philosophy

Chemical Engineering

Engineering and Physical Sciences

University of Birmingham

University of Birmingham Research Archive

e-theses repository This unpublished thesis/dissertation is copyright of the author and/or third parties. The intellectual property rights of the author or third parties in respect of this work are as defined by The Copyright Designs and Patents Act 1988 or as modified by any successor legislation. Any use made of information contained in this thesis/dissertation must be in accordance with that legislation and must be properly acknowledged. Further distribution or reproduction in any format is prohibited without the permission of the copyright holder.

ii

Rian, for helping me find the light in my day

iii

ii. Abstract

The work presented within this thesis focuses on the synthesis, characterisation and

processing of novel materials for use within solid oxide fuel cells.

A range of perovskite materials, previously shown to have potential for solid oxide fuel

cell applications, were selected for further studies. These included LaMnO3, SrFeO3-δ

and Sr0.8Ti0.6Nb0.4O3-δ. These materials were doped with various dopants, including

cations such as V5+

and Ti4+

and also SiO44-

oxyanions. Once doped, the materials were

analysed by X-ray powder diffraction and underwent testing to ascertain their suitability

for use as solid oxide fuel cell electrodes. This included identifying structural stability in

anode conditions alongside thermal expansion studies. Overall, improvements over

undoped samples were noted, especially for Sr0.8Ti0.6Nb0.4O3-δ samples doped with V5+

and SrFeO3-δ samples doped with SiO44-

, although LaMnO3 doped with Ti4+

proved less

successful.

Production methods for the formation of microtubular solid oxide fuel cells were also

investigated. Powder processing for paste formation was examined, for subsequent use in

extrusion. The extrusion process was also investigated, alongside debinding and sintering

studies. The development of a reliable and repeatable process for cell production proved

difficult, especially on a smaller scale necessary to facilitate the testing of novel

materials.

iv

iii. Acknowledgements

I would like to thank Prof. Peter Slater for all of his help, encouragement and instruction

for all of the work carried out under his tutelage. I would also like to extend my thanks to

Prof. Kevin Kendall, Prof. Robert Steinberger-Wilckens, Dr. Waldemar Bujalski, Dr.

Aman Dhir and Dr. Jose Porras for their input and assistance.

To all of the CDT members, I thank you for your company, help and distraction over the

past 4 years. I'd like to thank EPSRC for funding this work, and countless people for

helping me carry it out; Carl Meggs, Jackie Deans, Bob Sharpe and the rest of the

workshop staff, Artur Majewski, John Hooper, Lynn Draper and the rest of the office

staff, Cathryn Hancock, Scott Taylor and Prof. Stuart Blackburn amongst others.

I'd like to thank my family for their unwavering support and encouragement and for

pushing me to pursue a PhD. To Rian, you've been a rock through all of this, keeping me

motivated and not letting me give in or give up. To the pool players, I congratulate each

and every one of you for being just as abysmal after at least two years of practise as you

were when we started having a quick break - and continuing to blame poor skill on the

cleft of doom.

To the Slater group and the rest of floor 5, thanks for all of the assistance over the years.

Much of this work would have been impossible had it not been for your help.

v

iv. Table of Contents

i. Title ................................................................................................................................... i

ii. Abstract......................................................................................................................... iii

iii. Acknowledgements ...................................................................................................... iv

iv. Table of Contents ....................................................................................................... v

v. List of Figures ............................................................................................................. viii

vi. List of Tables ............................................................................................................... xii

vii. List of Equations........................................................................................................ xiv

viii. List of Abbreviations ................................................................................................ xvi

Chapter 1. Introduction ....................................................................................................... 1

1.1. Clean Energy Generation ......................................................................................... 1

1.2 Solar Power ............................................................................................................... 3

1.3. Wind Power.............................................................................................................. 3

1.4. Geothermal ............................................................................................................... 4

1.5. Fuel Cells ................................................................................................................. 5

1.5.1. Hydrogen as a Fuel ............................................................................................ 6

1.5.2. Methane as a Fuel .............................................................................................. 7

1.5.3. Fuel Cell Material Requirements ...................................................................... 8

1.5.4. Types of Fuel Cell ........................................................................................... 10

Chapter 2. Solid Oxide Fuel Cell Materials ...................................................................... 17

2.1. Anode Materials ..................................................................................................... 20

2.1.1. Current Material Limitations ........................................................................... 20

2.1.2. Novel Material Research ................................................................................. 25

2.2. Cathode Materials .................................................................................................. 33

2.3. Electrolyte Materials .............................................................................................. 35

2.4. Project Aims ........................................................................................................... 38

Chapter 3. Experimental Techniques ................................................................................ 39

3.1. Synthesis Techniques ............................................................................................. 39

3.1.1. Solid State Synthesis ....................................................................................... 39

3.1.2. Co-Precipitation............................................................................................... 40

vi

3.1.3. Sol-Gel............................................................................................................. 41

3.1.4. Hydrothermal synthesis ................................................................................... 41

3.2. Pellet Production for Conductivity and Dilatometry Experiments ........................ 43

3.3. X-Ray Diffraction (XRD) ...................................................................................... 44

3.4. Crystal Symmetry .................................................................................................. 47

3.5. Point Groups and Space Groups ............................................................................ 51

3.6. Rietveld Refinement .............................................................................................. 51

3.7. Thermogravimetric Analysis coupled to Mass Spectrometry (TGA-MS) ............. 54

3.8. Dilatometry ............................................................................................................ 55

3.9. Electrical Testing ................................................................................................... 57

3.9.1. Four Probe D. C. Conductivity ........................................................................ 58

3.9.2. Electrochemical Impedance Spectroscopy (EIS) ............................................ 59

3.9.3. Fuel Cell Testing ............................................................................................. 67

3.10. Microtube Manufacture ........................................................................................ 73

3.10.1. Powder Preparation ....................................................................................... 74

3.10.2. Paste Formation ............................................................................................. 75

3.10.3. Cold Extrusion ............................................................................................... 77

3.10.4. Hot Extrusion ................................................................................................ 82

3.10.5. Debinding and Sintering ................................................................................ 82

3.10.6. Layering......................................................................................................... 83

3.11. Scanning Electron Microscope (SEM) ................................................................ 87

Chapter 4. Extrusion and Fabrication of Microtubular SOFCs ........................................ 91

4.1. Introduction ............................................................................................................ 91

4.2. Experimental and Results....................................................................................... 93

4.2.1. Paste Formation ............................................................................................... 93

4.2.2. Extrusion ......................................................................................................... 95

4.2.3. Re-Machined Die Set Samples ...................................................................... 102

4.2.4. Hot Extrusion Samples .................................................................................. 102

4.2.5. Cell Testing ................................................................................................... 104

4.2.6. SEM Imaging ................................................................................................ 109

4.2.7. Further Extrusion Work................................................................................. 113

4.8. Further Work and Conclusions ............................................................................ 118

vii

Chapter 5. Synthesis and characterisation of Si doped SrFeO3-δ for both cathode and

anode applications .......................................................................................................... 119

5.1. Introduction .......................................................................................................... 119

5.2. Experimental ........................................................................................................ 121

5.3. Results and Discussion......................................................................................... 123

5.3.1. Synthesis and Structure Determination ......................................................... 123

5.3.2. Electrical Conductivity Measurements ......................................................... 127

5.3.3. Area Specific Resistance ............................................................................... 131

5.3.4. Dilatometry .................................................................................................... 135

5.3.5. Fuel Cell Testing ........................................................................................... 138

5.4. Conclusions .......................................................................................................... 140

Chapter 6. Synthesis and characterisation of V doped Sr0.8Ti0.6Nb0.4O3-δ for SOFC

Applications .................................................................................................................... 141

6.1. Introduction .......................................................................................................... 141

6.2. Experimental ........................................................................................................ 143



6.3.2. 4-Probe DC Conductivity .............................................................................. 151

6.3.3. Area Specific Resistance ............................................................................... 153

6.3.4. Reoxidation Studies ....................................................................................... 156

6.3.6. Dilatometry .................................................................................................... 158

6.4. Conclusions and Further Work ............................................................................ 161

Chapter 7. Synthesis and characterisation of La1-xMnxMn1-xTixO3-δ for SOFC

applications ..................................................................................................................... 162

7.1. Introduction .......................................................................................................... 162

7.2. Experimental ........................................................................................................ 164



7.3.2. Conductivity Measurements .......................................................................... 168

7.4. Conclusions .......................................................................................................... 173

Chapter 8. Conclusions and Further Work ..................................................................... 174

Chapter 9. Publications ................................................................................................... 178

Chapter 10. References ................................................................................................... 178

viii

v. List of Figures

Figure 1. Diagram of the most common types of fuel cell, the fuel and oxidising agents

required, typical operating temperatures and the electrolyte ion transfer type and

direction

Figure 2. Cubic and orthorhombic perovskite structures

Figure 3. An example of the triple phase boundary in a SOFC utilising a YSZ/NiO

cermet as the anode

Figure 4. Two crystal planes with spacing ‘d’ showing reflection of X-rays

Figure 5. The 14 Bravais lattice unit cell types [106]

Figure 6. Miller Indices for several lattice planes [106]

Figure 7. Diffraction pattern undergoing refinement using the GSAS suite

Figure 8. Electrode configuration for 4-probe DC conductivity

Figure 9. Example of a Nyquist-plot for ionic conducting solids, demonstrating different

physical phenomena. Z’’ = imaginary impedance, Z’ = real impedance

Figure 10. Schematic diagram of the fuel cell test equipment used in this thesis

Figure 11. Schematic diagram of the pellet cell test setup used in this thesis

Figure 12. Example fuel cell polarisation curve showing the three main sources of

voltage losses

Figure 13. Schematic of an anode-supported m-SOFC [111]

Figure 14. The Bridge twin roll mill used for this work

Figure 15. Diagram detailing a basic extrusion technique

Figure 16. Extrusion rod complete with mandrel, without and with extrusion die and

alignment cylinder

ix

Figure 17. Extrusion ram and mandrel with YSZ/NiO paste wrapped around it. (L) shows

before insertion into extrusion barrel, (R) is setup complete with die as it would be inside

the barrel

Figure 18. Instron 4467 with 30kN load cell and sample installed (L), close-up of sample

ready to extrude (R)

Figure 19. Flow diagram for microtubular solid oxide fuel cell manufacture

Figure 20. Graph showing sintering profiles for NiO-only samples (batch 1A) and CuO-

containing samples (batches 1B and 1C)

Figure 21. Voltages for cells moving from OCV to 0.05 A current demand


Figure 23. Voltages and currents for cells moving from OCV to 0.9 V voltage demand

Figure 24. SEM images of tubes extruded of various batches and in various states of

production. a = YSZ-NiO-CuO as extruded, b = YSZ-CuO debound and sintered at 1100

°C, c = YSZ-NiO debound and sintered at 1400 °C, d = YSZ-CuO as extruded, e = YSZ-

NiO debound and sintered at 1300 °C then YSZ electrolyte deposited and sintered at

1300 °C, f = YSZ-NiO debound and sintered at 1100 °C, g = YSZ-NiO as extruded

Figure 25. Voltages and currents for cells (extruded using the remanufactured die set)

moving from OCV to 0.9 V voltage demand

Figure 26. XRD pattern for SrFeO3-δ in air (bottom) and annealed in 5%H2/N2 at 600 °C

for 24 hours (top)

Figure 27. XRD pattern for SrFe0.9Si0.1O3-δ in air (bottom) and annealed in 5%H2/N2 at

600 °C for 24 hours (top)

Figure 28. Kröger-Vink notation showing silica incorporation causes oxygen vacancies

and Fe reduction

x

Figure 29. Plot of conductivity against temperature for the series SrFe1-xSixO3-δ in air

Figure 30. Plot of conductivity against temperature for the series SrFe1-xSixO3-δ in

5%H2/N2

Figure 31. Nyquist plot for SeFe0.9Si0.1O3- δ at 800°C in air, including fitting

Figure 32. Nyquist plot for SeFe0.9Si0.1O3- δ at 600°C in 5%H2/N2, including fitting

Figure 33. Dilatometry data for SrFeO3-δ from 20 - 1000 °C at 15 °C min-1

in air. Solid

line is TEC profile, dashed line is temperature profile

Figure 34. Dilatometry data for SrFe0.95Si0.05O3-δ from 20 - 1000 °C at 15 °C min-1

in air.

Solid line is TEC profile, dashed line is temperature profile


in air.


Figure 36. Voltages and currents for symmetrical pellet cells of SrFe0.9Si0.1O3-δ on YSZ

at 800°C moving from OCV to 0.85 V voltage demand

Figure 37. Graph showing unit cell volume (Å3) for Sr0.8Ti0.6Nb0.4-xVxO3-δ against V

content

Figure 38. XRD pattern for Sr0.8Ti0.6Nb0.2V0.2O3-δ in air (bottom) and annealed in

5%H2/N2 at 600 °C for 24 hours (top)



Figure 40. XRD pattern for Sr0.8Ti0.6V0.4O3-δ in air (bottom) and annealed in 5%H2/N2 at


Figure 41. XRD pattern for Sr0.75Ti0.5V0.5O3-δ in air showing single phase purity

Figure 42. Graph showing unit cell volume (Å3) for Sr0.8Ti0.6Nb0.4-xVxO3-δ (unreduced

and reduced at 930 °C) against V content

xi

Figure 43. Plot of conductivity against temperature for the series Sr0.8Ti0.6Nb0.4-xVxO3-δ

in 5%H2/N2

Figure 44. ASR Nyquist plot for Sr0.8Ti0.6Nb0.2V0.2O3-δ at 800°C in air, including fitting

Figure 45. ASR Nyquist plot for Sr0.8Ti0.6Nb0.2V0.2O3-δ at 800°C in 5%H2/N2, including

fitting

Figure 46. Conductivity against time for samples of Sr0.8Ti0.6Nb0.4-xVxO3-δ exposed to air

post reduction in 5%H2/N2

Figure 47. Dilatometry data for Sr0.8Ti0.6Nb0.4O3-δ from 20 - 1000 °C at 15 °C min-1

in

air. Solid line is TEC profile, dashed line is temperature profile

Figure 48. Dilatometry data for Sr0.8Ti0.6Nb0.2V0.2O3-δ from 20 - 1000 °C at 15 °C min-1

in air. Solid line is TEC profile, dashed line is temperature profile

Figure 49. XRD pattern for La0.8MnTi0.2O3-δ in air (bottom) and annealed in 5%H2/N2 at






Figure 52. Variation in the conductivities of the series La1-xMnxMn1-xTixO3-δ in air and

5%H2/N2 at 850 °C

Figure 53. Log10 σ vs 1000/T for the series La1-xMnxMn1-xTixO3-δ in air

Figure 54. Graph of Log10 σ vs 1000/T of series La1-xMnxMn1-xTixO3-δ in 5%H2/N2

xii

vi. List of Tables

Table 1. Unit cell symmetries, side lengths and angles

Table 2. Quantities of powders used to make up the cermet for different samples

Table 3. Binder materials used for paste formation

Table 4. Firing Profile for sample 1A

Table 5. Firing Profile for samples 1B and 1C

Table 6. Initial composition of electrolyte slurry used for dip coating

Table 7. Composition of cathode inks 1 and 2

Table 8. Composition of electrolyte slurry used for manual deposition

Table 9. Unit cell parameters and volumes (per formula unit) for the series SrFe1-xSixO3-δ

Table 10. Oxidation state and oxygen deficiencies for the series SrFe1-xSixO3-δ

Table 11. Conductivity data for SrFe1-xSixO3-δ in air

Table 12. Conductivity data for SrFe1-xSixO3-δ in 5%H2/N2

Table 13. ASR values for the series SrFe1-xSixO3-δ in air

Table 14. ASR values for the series SrFe1-xSixO3-δ in 5%H2/N2

Table 15. TEC data for the series SrFe1-xSixO3-δ

Table 16. Unit cell size and cell volume for Sr0.8Ti0.6Nb0.4-xVxO3-δ

Table 17. Unit cell size and cell volume for Sr0.8Ti0.6Nb0.4-xVxO3-δ reduced at 930 °C in

5%H2/N2

Table 18. 4-probe DC conductivity data for the series Sr0.8Ti0.6Nb0.4-xVxO3-δ in 5%H2/N2

post reduction at 930 °C

Table 19. ASR data for the material Sr0.8Ti0.6Nb0.2V0.2O3-δ in both air and 5%H2/N2

xiii

Table 20. Conductivity (S cm-1

) against time (s) for samples of Sr0.8Ti0.6Nb0.4-xVxO3-δ

exposed to air post reduction in 5%H2/N2

Table 21. TEC data for the materials Sr0.8Ti0.6Nb0.4O3-δ and Sr0.8Ti0.6Nb0.2V0.2O3-δ

Table 22. Unit cell parameters and cell volumes for the series La1-xMnxMn1-xTixO3

Table 23. Conductivity data in air for the series La1-xMnxMn1-xTixO3

Table 24. Conductivity data in 5%H2/N2 for the series La1-xMnxMn1-xTixO3 in 5%H2/N2

xiv

vii. List of Equations

Equation 1. Steam reformation of methane

Equation 2. Goldschmidt's tolerance factor calculation for perovskite stability (where t =

tolerance factor, rA/rB/rO = ionic radius of the A-site/B-site/O-site)

Equation 3. Sulphur poisoning of nickel within SOFC anodes

Equation 4. Partial oxidation of methane

Equation 5. Dry reformation of methane

Equation 6. Bragg's law, where n = integer, d = crystal plane spacing, λ = wavelength of

X-rays used and θ = angle of the X-rays in relation to the plane

Equation 7. Expression for calculation of linear thermal expansion coefficients, where a

= TEC, Li = initial length, Lf = final length, Ti = initial length and Tf = final length

Equation 8. Calculation for 4 probe DC conductivity (where σ = conductivity, L =

distance between electrodes 2 and 3, A = pellet cross-sectional area, I = current passed

between electrodes 1 and 4, V = voltage difference between electrodes 2 and 3)

Equation 9. Expression for a sinusoidal voltage, where e = observed voltage at time t, E

= input voltage and ω = frequency (in Hz) times 2π

Equation 10. Expression for the phase shift between input voltage and output current,

where i = current, I = the maximum amplitude of the current and φ = phase angle

between input and output

Equation 11. Expression for the capacitive resistance Xc for a pure capacitor, where Xc =

1/ωC

Equation 12. Expression for the current of a pure resistor, where R = resistance

xv

Equation 13. Expression for the voltage-current relationship for a pure capacitor, where

Ē = voltage amplitude, j = √-1 and Ī = current amplitude

Equation 14. Capacitance and resistance components in series

Equation 15. Expression defining impedance Z

Equation 16. Expression showing the two components of the impedance, where ZRe =

real part (R) and ZIm = imaginary part (Xc)

Equation 17. Expression for the calculation of activation energy as a function of the

properties of the sample and the temperature, where σ = conductivity, T = temperature,

A = sample surface area, Ea = activation energy and R = resistance

Equation 18. Typical impedance (Z) where Z’ is the real part, j is the imaginary unit, ω is

the frequency and Z’’ is the imaginary part

Equation 19. Calculation to determine the ASR from the intercept points on a Nyquist

plot, where IH = high intercept, IL = low intercept, n = number of sides (in the case of a

pellet, always 2) and A = area of one face

Equation 20. Gibbs free energy change of the reaction of hydrogen and oxygen to form

water

Equation 21. Gibbs free energy change for the reaction of hydrogen and oxygen to form

water, with respect to partial pressure of gases

Equation 22. Calculation of electromotive force between anode and cathode in a fully

reversible system

Equation 23. Nernst equation for the reaction of hydrogen and oxygen to form water

xvi

viii. List of Abbreviations

AEMFC Alkaline Exchange Membrane Fuel Cell

AFC Alkaline Fuel Cell

ASR Area Specific Resistance

BSCF Ba0.5Sr0.5Co0.8Fe0.2O3−δ

BZCY Ba(Zr/Ce)1-yYyO2-y

CGO Cerium Gadolinium Oxide

CVD Chemical Vapour Deposition

EDX Energy-Dispersive X-ray spectroscopy

EIS Electrochemical Impedance Spectroscopy

FIT Feed In Tariff

GDC (See CGO)

GSAS General Structure Analysis System

ICE Internal Combustion Engine

IT-SOFC Intermediate Temperature Solid Oxide Fuel Cell

LDC La0.4Ce0.6O2-δ

LNF LaNi1-xFexO3

LSCF (La/Sr)1-xCoyFe1-yO3-δ

LSCM (La/Sr)1-xCryMn1-yO3-δ

LSF La0.6Sr0.4FeO3-δ

LSGM La1-xSrxGa0.8Mg0.2O3-δ

LSM La1-xSrxMnO3-δ

MCFC Molten Carbonate Fuel Cell

MFC Mass Flow Controller

xvii

m-CHP micro Combined Heat and Power

MIEC Mixed Ionic-Electronic Conductor

m-SOFC microtubular Solid Oxide Fuel Cell

NASA North American Space Agency

NiS Nickel Sulphide

OCV Open Cell Voltage

PAFC Phosphoric Acid Fuel Cell

PEHD High Density Polyethylene

PEMFC Proton Exchange Membrane Fuel Cell

PFA Pore Forming Agent

PMMA Polymethylmethacrylate

PPM Parts Per Million

PVA Polyvinyl Alcohol

PVB Polyvinyl Butaryl

PVD Physical Vapour Deposition

SDC Samaria-Doped Ceria

SEM Scanning Electron Microscope

SOFC Solid Oxide Fuel Cell

STP Standard Temperature and Pressure (298K, 100kPa)

TEC Thermal Expansion Coefficient

TGA-MS Thermogravimetric Analysis - Mass Spectrometry

TPB Triple Phase Boundary

TPO Temperature Programmed Oxidation

XRD X-Ray Diffraction

YSZ Yttira-Stabilised Zirconia

1

Chapter 1. Introduction

This Chapter introduces the concept of clean energy generation and the need for its

implementation. It covers a broad view of the industry and relevant technologies, largely

centred on fuel cells. It finishes with an in-depth look at solid oxide fuel cells (SOFCs)

including the requirement for materials to be used within these systems.

1.1. Clean Energy Generation

The UK is committed to reducing carbon emissions by 60% relative to 1990 emissions

by 2050, a level which will be hard to achieve due to the continuing trend towards

increasing energy consumption. In order to be on track to achieve this goal, the target of

completely decarbonising the electricity system by 2030 has been identified as a key

stage towards achieving the reduction in CO2 emissions required. This can be achieved

by replacing the generation capacity of older power station with high carbon emissions

per gigawatt-hour (GWh), such as coal-fired plants, with renewable energy sources such

as wind farms, geothermal power plants, nuclear power and various solar-based

technologies. However, a major drawback preventing the large-scale uptake of many

forms of renewable energy is the unreliable nature of generation. For example, the wind

blows intermittently so power can be generated at times of low demand such as the

middle of the night, and conversely at times of high demand the wind may be calm

leading to low levels of power generation. This increases the difficulty of the task of grid

balancing, the act of ensuring that the demands for electricity on the grid are matched to

2

the output of all generators attached to the grid, which must be carried out in real time.

This is perceived to be a major obstacle to overcome before renewable generation can

become a large proportion of the energy mix; indeed, it has been estimated that once

renewables reach 30% of the total grid capacity the stability of the system will be

compromised without large-scale capacity for energy storage [1].

A concept that has come to the fore in recent years is that of microgeneration. In the UK,

this has largely been driven by feed-in tariffs (FITs) and is most noticeable by the uptake

of solar photovoltaic installations. However microgeneration, like the FITs, is not limited

to solar photovoltaic and covers a wide range of technologies such as small-scale wind

turbines and Micro Combined Heat and Power (micro-CHP). Microgeneration has the

advantage of having virtually no transmission losses as the electricity is usually

consumed close to the source. It is also useful to allow communities to continue

functioning if the national grid was to go down. However, the smaller scales involved

generally leads to lower efficiencies, offsetting the transmission losses. In addition, it

makes management of the national grid more difficult as the production sites are not

centralised and controllable in the way that large-scale power stations are. Nevertheless,

the lower average carbon footprint of power generated against traditional coal, gas or oil

fired power stations is still reason enough for many governments to provide incentives to

increase installations and capacity for microgeneration.

3

1.2 Solar Power

Solar power harnesses the energy given off by the sun. This can be achieved in several

ways; the most commonly seen is photovoltaics, and arguably the most simple is solar

thermal hot water panels. The former is the conversion of electromagnetic radiation to

electric current; while the latter involves the direct use of heat which is transported by a

fluid medium, usually water. However there are a number of other technologies that

make use of the sun’s energy, including turbines driven by a heated fluid medium such

as liquid sodium or supercritical steam [2]. Storage of this heated fluid can allow

generation to take place at times of low solar flux, such as overcast periods of the day, or

even at night, which can help to balance the load on the grid.

In the UK, the uptake of small-scale photovoltaic systems has greatly expanded since the

introduction of the FITs, to the point where the value of the tariffs was reduced in 2011

to reflect the increase in installations and reduction in purchase and installation cost.

1.3. Wind Power

Generating electricity from the power inherent in the wind generally involves the use of

tall wind turbines. These have a series of blades that are driven by the wind and drive an

alternator. The location of these turbines can vary but is greatly affected by land

topography as this has a direct influence on wind speeds. It is for this reason that many

banks of wind turbines are located in the seas and oceans close to the shoreline; this

leads to their name, offshore wind farms. There are advantages to situating wind farms

4

onshore, however, as construction costs are typically much lower than those offshore.

The World Wind Association has calculated that as of June 2013 the capacity of wind-

generated electricity has surpassed 296 GW [3]. With total generation capacity in the

region of 4900 GW, this represents 6% of worldwide power production, a huge increase

even on 2005 levels where capacity was around 50 GW.

It is also becoming increasingly common to fit small turbines onto or into tall structures

and within cities. These largely rely on winds being channelled between buildings to

provide a stronger air flow, but can also harness the higher wind speeds present above

the height of surrounding buildings. For example, the 'Strata' tower, located in the

Elephant and Castle district of London, has 3 turbines rated at 19 kW each incorporated

into its pinnacle which can produce up to 50 MWh of electricity every year [4].

1.4. Geothermal

Energy can also be extracted from bedrock where the temperature is great enough; this is

referred to as geothermal energy production. There are several different methods used to

extract the energy from the rock, each with their own advantages and disadvantages.

These include the use of naturally occurring steam to drive turbines, pumping water

down to extract the heat directly through powering a turbine, or by using a binary system

where the heated water goes through a heat exchanger and heats a fluid medium in a

closed system which drives a turbine. There is also the potential for heated water to be

directly supplied to houses for heating purposes. Each variation on geothermal

technologies is suited to certain geographical features and rock types and is reliant on

5

sufficient heat being present close enough to the earth’s surface to warrant extraction.

This generally hinges around a high temperature differential in the crust of the earth to

provide a large potential heat capacity, without having to pump the heated fluid too far

down into the ground. The geographic limitations of geothermal generation severely

curtail the expansion of this technology, meaning that while it remains a valid

contributor to a lower-carbon future it will likely struggle to be a primary source of

power worldwide and will remain a viable source of power in a small number of

locations only, such as Iceland and the Philippines [5].

1.5. Fuel Cells

Fuel cells work on a similar principal to batteries, namely they convert chemical energy

to electrical energy via electrochemical reactions. However, unlike a battery, a fuel cell

requires an external supply of fuel such as hydrogen. While this makes the system more

complex, it also dramatically increases the potential time that the cell can run for, as well

as removing the lengthy recharge time inherent to batteries. Fuel cells can also be scaled

in a similar way to batteries with a large range of power capacities possible, potentially

able to power small devices only requiring a few milliwatts of power through to large

scale applications generating megawatts of power. However, it should be stressed that

fuel cells are not directly a green energy generation technology, as they require fuel to

run and that fuel must be produced. They are most commonly linked with hydrogen,

which doesn’t exist naturally on earth as molecular hydrogen and must be obtained from

other sources, such as splitting water into elemental oxygen and hydrogen. However,

certain fuel cell systems such as SOFCs can theoretically run on a large range of fuels

6

including hydrocarbons, allowing them to bridge the technological gap between

hydrocarbon fuels currently widely available and those available in the future such as

hydrogen. One of the main advantages of fuel cell systems is that they are very efficient,

often reaching over 50% system efficiencies, so when compared to other technologies

such as the internal combustion engine (ICE), that typically reach peak efficiencies of

25%, the benefits can clearly be seen.

1.5.1. Hydrogen as a Fuel

The case for hydrogen as a fuel has many positive points. When hydrogen reacts with

oxygen, either through combustion or in an electrochemical cell, the only by-product is

water. Hydrogen therefore has zero harmful emissions at the point of use. Unfortunately,

elemental hydrogen is not present on earth and so hydrogen must be produced by other

means. There are several means to produce hydrogen but all require an energy input of

some kind; for example, electrolysis of water produces hydrogen and oxygen but

requires a large electrical input, negating the cleanliness of the fuel at the point of use

unless the energy input is also clean, e.g. from renewable energy sources.

The majority of the world's supply of hydrogen is currently captured as a by-product of

hydrocarbon processing. This accounts for approximately 95% of hydrogen

commercially available and clearly is not a suitable source for hydrogen if the purpose of

utilising hydrogen is to reduce the carbon footprint of your fuel. It is therefore important

to move towards renewable production of hydrogen.

7

In this respect, it may be possibly to use the unpredictable nature of most renewable

energy sources to the advantage of hydrogen production. This could be achieved through

utilising the electricity to power hydrogen production, for example electrolysis, when

there is excess being produced, and so use the production of hydrogen as a form of

energy storage [6]. This production could be carried out anywhere with an electricity

grid connection, but would need to be controlled as part of the grid in order to be most

effective.

1.5.2. Methane as a Fuel

As some fuel cells such as SOFCs can utilise hydrocarbons as fuels, another use for

excess grid electricity could be the production of gas such as methane, again to act as a

form of energy storage. Once again, production could be distributed and would not need

to be local to electricity production. However, producing methane has a distinct

advantage over the production of hydrogen, namely the natural gas grid network. This

network could accept all of the methane processed and so minimise the need for local

storage. If the methane was produced from renewable energy sources it could also

contribute towards lowering the carbon footprint of the gas grid, and reduce the

dependence on foreign fuel imports, so giving much greater energy security. This idea of

using renewably-produced methane lends support for the continuing development of

SOFCs since these have the fuel flexibility to take advantage of the gas grid while

remaining electrically efficient.

8

In order for methane or other hydrocarbons to be utilised as a fuel within most cells,

reformation must take place. This is most commonly carried out in the presence of water

and is known as steam reformation, the endothermic reaction of water with methane to

produce hydrogen and carbon monoxide, as seen in Equation 1.

CH4 + H2O ↔ CO + 3H2 (1)

Equation 1. Steam reformation of methane

Steam reformation of methane is necessary due to the requirement of hydrogen as a fuel

for most types of fuel cell, and typically must be carried out before the fuel enters the

cell. However, in the case of high temperature cells such as SOFCs the reformation can

be performed within the cell itself as the high temperatures provide the energy required

to drive the reaction towards hydrogen production.

1.5.3. Fuel Cell Material Requirements

A fuel cell is a device that converts chemical energy to electrical energy by way of an

electrochemical reaction, with heat as the main by-product. In the most basic analysis, a

fuel cell has three distinct sections, the electrolyte, the anode and the cathode. Fuel is

introduced on the anode side of the system, an oxygenating substance (typically oxygen

in the air) is introduced at the cathode side, and they are separated by the electrolyte

layer. This electrolyte layer allows ions to pass through it, with the type of ion

determined by the type of fuel cell. The electrolyte material is also an electronic insulator

9

and so forces electrons to move around an external circuit where they do work, providing

the electric output of the system.

1.5.3.1. Electrolyte Layer

The electrolyte layer sits between the anode and cathode layers, providing a boundary

between them. It has different requirements of it depending on which fuel cell system is

being analysed; however, certain general characteristics must be present. It should be an

electrical insulator, forcing electrons to move separately around the system in order to

generate an electrical potential. It should also be an ionic conductor; the type of ions

conducted and the temperature of maximum conductivity depends on the type of fuel

cell. The different types of fuel cell are named after their electrolyte material; for

example, Proton Exchange Membrane cells rely on an electrolyte layer which

conductions protons.

1.5.3.2. Anode Layer

The anode layer is generally where the fuel first contacts the cell. It must be stable under

highly reducing conditions and provide some level of ionic conduction to allow ion

transfer between the electrolyte and anode materials, as well as electronic conduction to

allow the movement of electrons. In addition, it should be chemically inert with respect

to the electrolyte material. It should also provide catalytic activity in order to enhance

reactions taking place at the surface. These reactions will be detailed in upcoming

sections.

10

1.5.3.3. Cathode Layer

This must be stable in oxidising environments and is directly in contact with the

oxidising agent, typically oxygen from the air. Similarly to the anode, it should provide

some level of ionic and electronic conductivity in order to allow the movement of ions

between cathode and electrolyte and the movement of electrons, as well as also not

reacting with the electrolyte.

1.5.4. Types of Fuel Cell

There are several types of fuel cell under development, all of which have different

characteristics and potential uses. The most commonly studied are discussed below.

Figure 1 shows a breakdown of the different fuel cell types, their operating temperatures

and electrolyte types.

11

Figure 1. Diagram of the most common types of fuel cell, the fuel and oxidising agents

required, typical operating temperatures and the electrolyte ion transfer type and

direction

1.5.4.1. Proton Exchange Membrane Fuel Cell (PEMFC)

Proton exchange membrane fuel cells, also known as polymer electrolyte membrane fuel

cells, require high purity hydrogen to operate. They typically utilise carbon-supported

electrodes containing noble metals with an electrolyte consisting of proton-conducting

polymers such as Nafion®. Hydrogen enters the cell on the anode side and is broken

down into protons, which flow through the electrolyte membrane, and electrons, which

flow around the external circuit to the cathode. Here, the protons and electrons react with

oxygen to form water as the only waste product. The high cost of the materials currently

12

used has led to much research focusing on alternatives to make these cells more

financially viable. In addition, the metal catalysts used are susceptible to poisoning from

contaminants present in the fuel supply. This stringent requirement for fuel purity mean

that reformers intended to supply the cell with hydrogen from a hydrocarbon fuel source

must have additional filters, or be more complex than if the cell was less susceptible to

poisoning.

Operating temperatures are typically in the region of 50 - 150 °C, so PEMFCs are

considered low temperature fuel cells. This low temperature requirement allows

PEMFCs to start and stop operation rapidly, making them highly suited to mobile

applications. However, it also means that they generally are unsuitable for combined heat

and power applications as the heat output is too low.

1.5.4.2. Alkaline Fuel Cell (AFC)

Alkaline fuel cells are amongst the longest-proven fuel cells, having been used in the

Apollo series of missions run by NASA. They are based around an electrolyte which is

strongly alkaline, normally concentrated sodium or potassium hydroxide in liquid form

with an inert porous solid support. Oxygen and water combine at the cathode to form

hydroxide ions, drawing electrons from the external circuit. The hydroxide ions pass

through the electrolyte substance and react with hydrogen at the anode to form water,

releasing electrons which can flow around the external circuit back to the cathode. AFCs,

like PEMFCs, are classified as low temperature fuel cells, with typical operating

temperatures between 25 – 75oC.

13

Carbon dioxide must be removed from the gas coming into the cathode side as it poisons

the cell by reacting with NaOH/KOH forming Na2CO3/K2CO3 which can block the

catalytic sites. As with PEMFCs, AFCs require high purity hydrogen. This gas

processing at both the anode and cathode sides drive up system complexity and increases

parasitic load, leading to lower overall system efficiencies. In addition, noble metals such

as platinum are commonly used within the electrodes which also drive up cost. This,

combined with the hazardous liquid electrolyte material used, is the reason that research

into and demand for alkaline fuel cells has waned somewhat, with more interest being

paid to PEMFCs. However, research is ongoing into cheaper, non-precious metal catalyst

materials, which are believed possible due to the more favourable electrode processes in

alkaline conditions, has recently reinvigorated research into AFCs.

1.5.4.3. Alkaline Exchange Membrane Fuel Cell (AEMFC)

Alkaline exchange membrane fuel cells are a relatively recent development, having first

been discussed at length in 2005 [7]. These cells use a solid alkaline polymer as the

electrolyte material which can tolerate operation in the presence of CO2, therefore

removing the requirement of cathode gas cleaning present in AFCs. The other

advantages of AFCs are still present, such as the potential for non-precious metal low

cost catalysts and low operating temperature. However, as with any new technology

there are still many obstacles to be overcome, such as the development of an effective

alkaline ionomer which is easily dispersed and maximises the ionic contact between the

ion membranes and electrode layers.

14

1.5.4.4. Phosphoric Acid Fuel Cell (PAFC)

Utilising an electrolyte composed of phosphoric acid at high concentrations, these cells

are much more tolerant to impurities than both PEMFCs and AFCs. The phosphoric acid

is typically used at such high concentrations that it forms pyrophosphoric acid (H4P2O7),

which freezes at room temperature; it is for this reason that the elevated operating

temperatures (150 - 200 °C) are usually maintained for the life of the cell, making them

ideally suited for uninterruptable power supply or continuous load applications. They are

also suitable for use in vehicles that spend the majority, if not all of their time, requiring

a power output, such as buses or ships.

Operation reactions are very similar to those in PEMFCs, however the electrolyte is a

configuration more similar to AFCs, with a liquid conducting medium supported by a

solid porous structure. However, similarly to AFCs, the hazardous electrolyte material

and also the constant operation requirement make maintenance difficult, and the

electrodes again typically also contain platinum, which drives the cost up.

1.5.4.5. Molten Carbonate Fuel Cell (MCFC)

Molten Carbonate Fuel Cells operate at a range of temperatures above 500oC, allowing

non-precious metals to act as catalysts in order to keep costs down. The electrolyte is

composed of a molten salt mixture which conducts carbonate ions, again within a porous

inert solid support. Carbon dioxide must be present at the cathode in order to generate

the CO32-

, and this is often achieved through recycling the exhaust gas to increase the

15

CO2 partial pressure. MCFCs can operate on a wide range on fuels, including some that

are poisons to other fuel cells, such as carbon monoxide. If running on hydrocarbons

then internal reforming can occur, allowing the system to utilise them directly. This fuel

flexibility grants the system a considerable advantage over many other fuel cells covered

previously, as MCFCs could be run on widely available commercial fuels without the

need for a change of infrastructure.

However, the high temperatures and hazardous electrolyte materials used do have their

drawbacks. In this respect the durability of the cell is compromised due to the corrosion

of components in the harsh chemical conditions. Significant current research is therefore

concentrated on materials which can withstand this environment while maintaining high

levels of efficiency. In addition, the high operating temperatures mean that a fast on-off

cycle is difficult to achieve, rendering MCFCs unsuitable to a wide range of applications

requiring temporary or intermittent power.

1.5.4.6. Solid Oxide Fuel Cell (SOFC)

Solid Oxide Fuel Cells have the highest operating temperature region, between 500 –

1000 °C. They make use of a solid gas-tight electrolyte, usually a ceramic, which can

conduct oxide ions. Oxygen enters the cell at the cathode side and is reduced to oxide

ions, which pass through the electrolyte layer. On reaching the anode they react with the

fuel, releasing the electrons they gained on the cathode side, which can flow back around

the external circuit. Developments have also been made for electrolyte materials which

conduct protons in addition to oxide ions, such as Ba(Zr/Ce)1-yYyO2-y (BZCY) [8];

16

however, this work is relatively new and requires considerably more research before a

commercially viable system could be produced.

SOFCs can operate at high efficiencies, in the region of 50%, on a variety of fuels. A

significant increase in this efficiency can be reached if the waste heat is also utilised.

This makes SOFCs ideal for combining with turbines, or for use in Combined Heat and

Power (CHP) systems. As for MCFCs, SOFCs are normally stable in the presence of

carbon monoxide and dioxide and do not need to use precious metals as catalysts

because of the high operating temperatures. Another similarity to MCFCs is that they can

run on a variety of fuels and in most cases can internally reform hydrocarbons. However,

there are still many issues to overcome to increase the performance and lifetime of

SOFCs. Nickel is often used in the anode as a catalyst and electron conductor, and

operates well on pure hydrogen, although when using hydrocarbons as the fuel it has

drawbacks. Sulphur is present in most hydrocarbon fuels and this irreversibly binds with

the nickel, reducing both catalytic activity and electronic conductivity in the anode,

severely affecting performance. In addition, nickel is a catalyst towards the formation of

carbon-carbon bonds which can lead to ‘coking’ at the anode. Coking involves the build-

up of carbon on the surface of the anode, blocking the catalytically active sites and

disrupting gas transfer, which significantly decreases cell performance, and will be

discussed in further detail in Chapter 2

This thesis addresses the material aspects of SOFC systems, concentrating primarily on

those which are exposed to reducing environments. The different materials used with

SOFCs will also be discussed in Chapter 2.

17

Chapter 2. Solid Oxide Fuel Cell Materials

There is much ongoing research into materials with potential applications in SOFCs. All

materials used within SOFC systems must fulfil certain conditions; for example, the

thermal expansion coefficients (TECs) for all materials within a single cell must be close

in value, otherwise upon heating different layers would expand at different rates and

cause cracking. In addition, chemical compatibility with other components and physical

stability at the high operating temperatures is required. The current industry standards for

SOFCs along with newer materials with the potential to replace the conventional ones

will be discussed here. The aim for this project is to expand knowledge and

understanding of current materials alongside the development of newer materials, in

order to deliver the potential to overcome many of the issues detailed earlier in this

thesis.

Much of the work presented within this thesis focuses on materials with the perovskite

structure, general formula ABO3-x, where A = larger cation, B = smaller cation. This

structure was first discovered as adopted by CaTiO3 by Gustav Rose in 1839, and was

named after the Russian scientist Lev Aleksevich von Perovski. The crystal structure was

first described by Victor Moritz Goldschmidt in 1926, at which time he also determined

an equation for calculating the structural stability of a perovskite material from the

atomic radii of the ions involved [9]. This tolerance factor equation can be seen in

Equation 2. The structure can be described as consisting of BO6 octahedra arranged in a

corner-sharing arrangement with an A cation situated in the 12-coordination site in the

18

centre of a cube of 8 such octahedra. Examples of perovskite structures can be seen in

Figure 2.

(2)

Equation 2. Goldschmidt's tolerance factor calculation for perovskite stability (where t =

tolerance factor, rA/rB/rO = ionic radius of the A-site/B-site/O-site)

Cubic Orthorhombic

Figure 2. Cubic and orthorhombic perovskite structures

The Goldschmidt's tolerance factor allows the prediction of the strain on the unit cell and

can also be used to make predictions about the symmetry of the system. An ideal

perovskite typically has a large A-site cation paired with a small B-site cation which

19

gives a tolerance factor as close to 1 as possible, and typically results in a cubic unit cell

[10]. A tolerance factor of less than 1, which corresponds to either a smaller A-site cation

or a larger B-site cation, typically leads to the need for the octahedra to tilt to aid

stability. This often leads to an lower symmetry unit cell, such as orthorhombic, with a

value for t of 0.8 - 0.9. Alternatively, a tolerance factor over 1 might be observed, which

is caused by too great a difference in cation size between A-site and B-site, in this case a

hexagonal unit cell may result.

In Figure 2 it can clearly be seen that the change in unit cell from cubic to orthorhombic

is due to the tilting of the BO6 octahedra. This is caused by having an A-site cation

which is too small relative to the unit cell size and so the octahedra tilt, which brings

additional anions into contact with the A-site cations. The degree of tilting is directly

related to the ratio of the size of the A and B site cations as determined by Equation 2,

with a tolerance factor closer to 1 displaying a lower degree of tilting.

Since the perovskite unit cell symmetry is highly dependent on the size relationship

between the A-site and B-site cations, any changes that may occur through oxidation or

reduction may bring about a change in symmetry, leading to a new material with

potentially radically different properties. A good example of the different possible phases

available for a given family of materials is that of SrFeO3-δ where 0 ≤ δ ≤ 0.5. SrFeO3

has the perovskite structure with tetragonal symmetry; however; upon heating in a

reducing atmosphere it is possible to form the material SrFeO2.875 (or alternatively

Sr8Fe8O23). In this structure, iron is present as both Fe3+

and Fe4+

and the oxygen

vacancies order so that a regular pattern of both FeO6 octahedra and FeO5 square

20

pyramids form. If reduced further a brownmillerite-type phase, SrFeO2.5 (or Sr2Fe2O5),

can be formed in which there are alternate layers of octahedral and tetrahedral Fe [11].

There is also evidence of the formation of other phases based around strontium ferrate,

such as SrFeO2.75 (Sr4Fe4O11) [12].

2.1. Anode Materials

This thesis focuses mainly on the synthesis and characterisation of materials for use as an

anode in SOFCs. The anode of a SOFC is where the electrochemical reaction between

the fuel and oxide ions occurs, with the additional strain on an anode-supported fuel cell

system of having to provide the mechanical strength as well. The properties must

therefore include: high electronic conductivity, good mechanical strength,

electrocatalytic activity towards a range of fuels including hydrogen and/or

hydrocarbons, tolerance to fuel impurities (such as sulphur); high surface area to allow

maximum power densities, stability in reducing atmospheres, and chemical and

mechanical compatibility with the electrolyte material used within the cell.

2.1.1. Current Material Limitations

Current commercial anode technologies focus mainly on cermets, a dispersion of metal

particles over a matrix of a ceramic. The ceramic is typically the same material as is used

in the electrolyte layer, which allows a match in thermal expansion coefficients (TEC) as

well as for the layers to bind together. It has been commonplace for many years in high

21

temperature cells for this cermet to consist of YSZ with nickel nanoparticles dispersed

over the surface [13]. The nickel serves to conduct electrons as well as act as a catalyst

towards the reaction of hydrogen and hydrocarbons with oxide ions. Between the nickel

and YSZ, many of the criteria for an anode material are met [14]. The cermet is typically

between 40-60% nickel oxide on manufacture, which reduces to nickel metal under

operating conditions.

A cermet contains, by definition, a mixture of metal and ceramic particles. However, as

it is uncommon for ceramics to conduct electrons or for elemental metals to conduct

ionic species, the ability to fully function as an anode is restricted to those areas where

the metals and ceramic are in contact together with fuel. This area is known as the triple

phase boundary (TPB), as it involves the interaction of the gases with a solid that

transports the ions and a solid that transports the electrons.

As can be seen in Figure 3, the triple phase boundary is the point at which the smaller

nickel and larger YSZ particles are touching which is also open to a porous gas channel.

As the electrochemical reactions taking place require the presence of electrons, oxide

ions and the gaseous fuel, with nickel transporting the electrons and YSZ transporting

oxygen ions, these boundaries must be within pores or channels in the material to allow

the fuel to come into contact with the ions. This leads to the need for a large surface area

combined with the nickel being very well dispersed across the surface of the YSZ

ceramic support [15]. However, if the nickel is too widely dispersed then there is little

interconnectivity between particles and this leads to a lowering of cell performance due

to the raised resistance to electron transport of the anode.

22

Figure 3. An example of the triple phase boundary in a SOFC utilising a YSZ/NiO

cermet as the anode

The increase in size of the TPB would be highly desirable as it would increase the

available surface for reactions to take place on. This would subsequently have the

potential to increase the performance of the cell. However, with investigations into

mixed ionic-electronic conductor (MIEC) materials, particle size and porosity become

important factors due to not needing a mixed material anode. MIEC material research

will be covered in greater depth later in this Chapter.

These nickel/YSZ cermets are typically produced through mixing of nickel oxide with

YSZ, either as powders or a slurry. Once the SOFC has been fabricated, the NiO is

typically reduced in-situ to the metal as the anode material is exposed to a reducing

environment. This also leads to an increase in porosity as this removal of oxygen causes

a reduction in density of the material.

23

However, while these Ni/YSZ cermets have many benefits, they also have many

drawbacks which has led to a large wealth of research to be carried out into alternative

materials. Most readily accessible hydrocarbons, even after they have been cleaned to

remove contaminants or molecules outside of the molecular weight range, still contain

traces of pollutants, the most common of which is sulphur in the form H2S. Ni-

containing cermets can be severely degraded through the use of sulphur-containing fuels

due to sulphur poisoning of nickel. The mechanism by which sulphur reacts with nickel

and forms an inactive Ni-S species is shown in Equation 3.

Ni + H2S(g) ↔ Ni + H2S(ads) ↔ NiS + H2(g) (3)

Equation 3. Sulphur poisoning of nickel within SOFC anodes

Studies have shown that at a concentration as low as 0.05 ppm at a typical SOFC

operating temperature of 750oC, YSZ/Ni anodes can be degraded by sulphur, which will

lead to a dramatic loss of performance over time, with the effectively-irreversible

formation of NiS [16, 17]. Sulphur is typically present in the range of a few ppm in

natural gas as delivered to most homes within the UK currently. It is often in the form of

methanethiol (CH3SH) or tetrahydrothiophene (C2H4S) which are often added to give the

gas supply a smell; methane has no odour otherwise so gas leaks would be very difficult

to detect. If the problem of anode poisoning by hydrogen sulphide can be overcome then

it will allow SOFCs to be much more easily integrated into the current energy

infrastructure.

24

The drop in performance when exposed to sulphur-containing fuels is not enough by

itself to justify such a large quantity of research into anode materials for SOFC.

However, in addition to the issues surrounding sulphur, nickel acts as a catalyst towards

the formation of carbon-carbon bonds; in fact, it is a widely used catalyst in industry for

hydrogenation and reforming. This formation of carbon-carbon bonds causes 'coking' of

cells when they are run on hydrocarbons. This coking is a deposition of carbon on the

surface of the anode and is particularly problematic when there is little steam present, as

the steam acts to oxidise the carbon deposits and so help to remove it. This coking

significantly lowers the performance of the cell by reducing the surface area available for

reactions to occur [18–20]. Equation 1, Equation 4 and Equation 5 cover both steam and

dry reformation and also partial oxidation of methane, the simplest hydrocarbon fuel.

CH4 + ½ O2 2H2 + CO (4)

Equation 4. Partial oxidation of methane

CH4 + CO2 2H2 + 2CO (5)

Equation 5. Dry reformation of methane

In addition, formation of carbon deposits on the anode can cause damage to the

microstructure of the anode. This may cause the delamination of materials or the

formation of microcracks, leading to poorer material interconnections and lowering cell

performance. It is possible to determine the nature of the carbon deposited on the cell

through temperature programmed oxidation (TPO). This involves subjecting the coked

cell to an oxygenating atmosphere and measuring the levels of carbon monoxide and

25

dioxide being produced (using a mass spectrometer), while raising in temperature.

Carbon which is only adsorbed onto the surface (amorphous) will oxidise around 300 °C,

whereas graphitic carbon will oxidise at 600 °C [19–22].

The processes involved in both carbon deposition and sulphur poisoning are varied.

Layers of both carbon and sulphur have been observed to grow on anode surfaces in a

similar fashion as chemical vapour deposition (CVD), for example [23]. Several methods

used for reducing poisoning effects (such as increasing steam content at the anodes to

allow steam reformation of methane) have significant unwanted side effects (such as

difficulties in thermal management due to the endothermic reforming process), further

complicating matters. Therefore, much work is ongoing into alternative materials which

do not suffer from carbon deposition or sulphur poisoning.

2.1.2. Novel Material Research

Significant research in recent years has focussed on replacing, either fully or to some

degree, nickel with copper within cermet systems [24]. Copper has been shown to have

the capability to reduce coking when running on hydrocarbons, even when the fuel is dry

[25, 26]. However, copper oxide (CuO) has a melting point of 1083 °C, which is far

lower than that typically used for the sintering of YSZ. This means that in order to use

copper as a catalyst within the anode of an anode-supported SOFC it must either form a

part of a cermet which contains a ceramic other than YSZ, which sinters at a temperature

lower than 1083 °C, or it must be added to the anode layer once all stages of fabrication

which involve temperatures over 1083 °C have been completed. In addition, Cu has a

26

lower catalytic activity towards the breaking of carbon-carbon and carbon-hydrogen

bonds than that of Ni. This lower catalytic activity is the reason behind copper being

more resistant to coking; however, it is problematic for operation on hydrocarbons as the

efficiency of internal reforming is significantly reduced. This has led to the use of CGO

in conjunction with Cu as this also displays some catalytic activity towards the

breakdown of hydrocarbons and can be sintered at lower temperatures than YSZ,

although this is still problematic [27, 28]. In addition, it has been shown in some recent

studies that both YSZ and CGO may be unstable in the presence of copper, leading to a

loss in oxide ion conductivity [29].

There is also a great deal of interest in materials based around CGO or utilising a cermet

or mixture of CGO and other materials, due to its catalytic activity towards the

breakdown of hydrocarbons and the fact that it is a better oxide ion conductor than YSZ

[30, 31]. Within the anode, the problems of electronic conduction above 600 °C are no

longer an issue as increased electronic conductivity is favourable. Unfortunately CGO is

in the region of 3 - 4 times more expensive than YSZ, reducing its appeal.

Investigations into MIEC materials have also been carried out, with many of these being

based around the perovskite structure. These typically feature transition metal elements

on the B site, usually first row transition elements due to the small size requirements of

the site. The use of MIEC materials within the electrodes of SOFCs would potentially

bring about a number of improvements over a Ni/YSZ cermet. The simplification of the

fabrication process would be advantageous, as additional processing of the nickel oxide

and YSZ would be required over a single material, and also because the nickel oxide

27

used is a suspected carcinogen and so additional safety precautions are required. The

conduction of both oxide ions and electrons within MIECs also aids the available area

for reactions to occur; the triple phase boundary is extended to any surface of the MIEC

material within the porous anode as O2-

can be supplied and electrons removed through

the same material. MIEC materials also typically don't suffer from particle

agglomeration to the same degree as a NiO system; nickel particles move over the

surface of the YSZ (or other oxide ion conductor material) framework over time, leading

to reduced surface area and consequently reduced cell performance [32, 33].

In terms of such systems, there is a lot of work focussing on perovskite titanates based

around SrTiO3 which has been known to have potential as an anode material [34]. This

compound provides redox stability, even at the high operating temperatures of a SOFC;

however, it is very difficult to reduce and this irreducibility also brings with it low

electronic conductivity. This can be countered through doping, both on the Sr and Ti

sites. However, many of these doped systems struggle with relatively low oxide ion

conductivity, in part due to the lack of oxide vacancies even after reduction. Many

different variants of this material have been made, with dopants such as Ba, Ca, La, Ni,

Mg, Mn, Fe, Cr, In, Sn, Nb [35–39]. However, they largely suffer from the same issue of

low oxide ion vacancies and hence oxide ion conductivity. In an attempt to counter this,

perovskite materials with A-site deficiencies have been synthesised, for example Sr1-

x/2Ti1-xNbxO3-δ [40–42]. These materials also display increased resistance towards

forming Ruddlesden-Popper type impurities (which reduce electrical conductivities) as

forming these layers requires an excess of A-site cations [43, 44].

28

Compounds based around the tungsten bronze structure of AxBO3, where x < 1, A is

usually an alkali or alkaline earth metal cation and B is a large transition metal cation,

such as W or Nb, have also been investigated. These are closely related to perovskite

systems and incorporate the same BX6 octahedra with A-site cations occupying

interstitial sites. Many variations can be synthesised, such as Sr0.6Ti0.2Nb0.8O3-δ and

Na0.8NbyW1-yO3-δ, which can display promising electronic conductivities [45–48].

However, slow oxygen ion transport caused by a diffusion-limited process, combined

with possible reactivity with YSZ are problems yet to be overcome.

The synthesis and analysis of A-site deficient perovskite materials, such as (LaxSr1-x)1-

yCr0.5Mn0.5O3-δ with y ≤ 0.1 has shown promise, with good stability shown at high

temperature in both oxidising and reducing environments [49, 50]. This series of

materials, however, suffers from similar issues to those of the titanates, namely low

oxide ion conductivity. In addition, catalytic activity is also low, necessitating the

inclusion of additional catalytically active materials. It has been shown that adding 4%

Ni to La0.8Sr0.2Cr0.8Mn0.2O3-δ/CGO increases cell performance when fuelled by

hydrocarbons, without significantly increasing carbon deposition [51]. This lack of

increase in coking is thought to be due to the low levels of Ni used. In addition, CeO2 has

been used as an additive with reasonable success [52]. This also displays a better

sustained performance when sulphur is present than those with the nickel additive.

Focus has also been placed on the series Sr1-yLayTiO3+δ [35, 53]. This has been shown to

display high electronic conductivities, as high as 100 - 500 S cm-1

between 700 - 1000 °C

[54]. However, only low oxide ion conductivities have been achieved without relying on

29

doping. Mn, Ga, Y, Al, Mg, Fe, Ce and Sc have all been doped onto the B-site and have

been shown to increase oxide ion conductivity at the expense of electronic conductivity

[54–56]. Study into A-site deficient analogues were been carried out and was also shown

to increase the ionic conductivity [41]. More work is required on these systems if they

are to be utilised as a composite anode, for example with YSZ, as the electronic

conductivity alone is too low to produce an efficient cell.

Other perovskites such as Sr1-xLaxVO3 have also been investigated. These vanadate

perovskites display very high electronic conductivities at higher temperatures [57].

Unfortunately, most are unstable in an oxidising environment, and this coupled with a

large volume change on reduction leads to incompatibility with many electrolytes, both

conventional and novel, as well as manufacturing difficulties. Mo doped compounds

such as CaV0.5Mo0.5O3-δ have been shown to have oxidised (a scheelite-type structure)

and reduced structures with very similar TECs in addition to high electronic conductivity

of over 500 S cm-1

at 800 °C [58]. Degradation of the material, however, was observed

above 600 °C, which may be a result of the presence of calcium on the A-site, as this has

previously been observed to increase electronic conductivity at the expense of redox

stability [59].

Research on Mo containing anodes has been built upon with studies into the double

perovskite Sr2MgMoO6 and related materials [60–62]. These have the potential to be

sulphur tolerant anodes with both high oxide ion and electronic conductivity. They have

also been shown to be stable in both oxidising and reducing environments, along with

good tolerance to the presence of S. Mo has, however, been shown in some cases to

30

diffuse into the YSZ electrolyte of the cell, which is detrimental to performance. To

counteract this issue, it has been shown that a CeO2 barrier layer can be successfully

employed to prevent this diffusion, albeit at the expense of additional cost and

complexity of the cell. The promising results of these molybdates is generating increased

interest into this material group.

Oxyanion doping into the perovskite structure is a doping strategy which has recently

begun to have a more rigorous examination. Whereas many traditional doping strategies

focus on replacing cations with aliovalent alternatives, incorporating oxyanions such as

nitrate, sulphate or phosphate has been shown to be possible on the B site within

perovskite systems. Through oxyanion doping it is possible to stabilise phases which

otherwise would not be stable; for example, Ba2In2-xPxO5+x (0.1 ≤ x ≤ 0.3) has increasing

cubic structure stability at lower temperatures on increasing levels of phosphate

concentration [63–65]. The orthorhombic crystal structure has ordered oxide ion

vacancies, as opposed to the disordered vacancies present in cubic structures; typically,

ionic conductivity increases with increasing oxide ion disordering. This stabilising of

phases can be seen in other ionic conducting solid, such as YSZ, where undoped zirconia

undergoes a phase change at a temperature too high to be of use.

The doping of oxyanions onto the B-site of perovskites is also being investigated for use

within electrodes of SOFCs. Initial studies based around SrCoO3-δ were carried out,

involving doping with phosphate, sulphate and silicate [66, 67]. Doping with between

3% - 7% of oxyanion onto the Co site caused a phase change from hexagonal to cubic,

subsequently leading to large increases in conductivity. However, further studies on

31

long-term stability showed that annealing above 600°C caused a gradual phase change

back to a hexagonal cell.

Similar investigations into the doping behaviour of the related hexagonal perovskite

SrMnO3 have also been carried out [68]. Higher levels of dopant were found to be

required at around 15%, with only silicate proving possible to stabilise the cubic phase.

The incorporation of the SiO44-

anion, which is tetrahedral, in both Co/Mn perovskites at

the expense of the B-site cation, which is octahedral, necessitates the formation of oxide

ion vacancies, leading to partial reduction of the remaining Co/Mn.

This allows doping with oxyanions to be used as a potential strategy to modify the

oxidation states of transition metal cations through oxide ion vacancy creation. This has

been demonstrated to be an effective means of improving electrical conductivity, even

when no change in cell symmetry is recorded, in the work carried out by Porras et al into

doping CaMnO3. In addition, doping with borate was shown to increase conductivity,

despite being of lower valence than the Mn it was replacing, which was again attributed

to the increase in electron doping through increased oxide ion vacancies outweighing the

expected decrease of the smaller charge species. It was not clear whether the the boron

was present as BO33-

or BO44-

, with further work being required in order to determine the

coordination in-situ.

Iron-containing perovskite systems, such as SrFe0.75Mo0.25O3-y have been investigated in

depth [69, 70]. Powder neutron diffraction studies show a tetragonal unit cell, and

Mössbauer and XAS studies show that oxidation states are +6 for Mo and +3/+4 for Fe.

32

This material showed promising regenerative properties to both sulphur poisoning and

coking, with the material regenerating on exposure to oxygen at operating conditions.

Mo content increases have also been investigated in the series SrMo1-xFexO3-y (with x ≤

0.3) [71, 72]. Fe was doped onto the B site of SrMoO3 in an attempt to introduce oxygen

vacancies and improve the oxide ion conduction, without compromising the high

electronic conduction of the parent material. SrMo1-xFexO3-y showed promising results

when utilised as the anode as part of a SOFC cell (with La0.8Sr0.2Ga0.83Mg0.17O3-δ

(LSGM) as the electrolyte and SrCo0.8Fe0.2O3 (SCFO) as the cathode), achieving

maximum power densities of 870 and 790 mW cm-2

at 850 °C on H2 (for x = 0.1 and 0.2

respectively). The thermal expansion coefficients correspond to many widely-used

electrolyte materials (in the region of 11 - 14 x 10-6

K-1

) and chemical compatibility with

LSGM was demonstrated.

Studies into another iron containing system, LaSr2Fe2CrO9-δ (La0.33Sr0.67Fe0.67Cr0.33O3-δ)

were undertaken. Cells were fabricated using LaSr2Fe2CrO9-δ/CGO anode, LSGM

electrolyte and LSCF cathode, with a La0.4Ce0.6O2-δ (LDC) barrier layer. A maximum

current density of ~400 mW cm-2

was recorded at 800 °C on wet hydrogen, and

resistance to sulphur poisoning was displayed at 22 ppm sulphur content. Good redox

performance was also noted; however, poor performance on methane indicates a lack of

catalytic activity.

33

2.2. Cathode Materials

The requirement for cathodic materials includes those standard to any fuel cell material

such as stability at high temperatures and matched TECs, in addition to catalytic activity

towards the dissociation of oxygen and electronic conductivity within oxidising

conditions. Many of these criteria are met by LaMnO3±δ and other perovskite systems

containing transition metals, which are widely used in SOFCs [34, 73, 74]. The high

electronic conductivity exhibited is a result of the mixed valency of the manganese,

present as both Mn3+

and Mn4+

. This is further enhanced through doping with alkaline

earth cations, most typically strontium, to form La1-xSrxMnO3-δ (LSM), which shows

very high electronic conductivity at the high operating temperatures of SOFC systems

[75]. Unfortunately, LSM exhibits relatively low oxide ion conductivity and has been

shown to react with YSZ, a common electrolyte material, at elevated temperatures,

which leads to an insulating layer preventing oxide ion transportation, reducing the

available triple phase boundary area and lowering the performance of the cell [76]. Other

related perovskite systems such as La0.6Sr0.4FeO3-δ (LSF) have also been studied in

detail, and were found to have favourable ionic and electronic conductivity alongside

good thermal expansion and chemical compatibility with both YSZ and CGO [77, 78].

Further work on these systems has been carried out leading to interest in (La/Sr)1-

xCryMn1-yO3-δ (LSCM) and (La/Sr)1-xCoyFe1-yO3-δ (LSCF) with x ≤ 0.1 for use at high and

intermediate temperatures respectively [79–81].

Research into potential new cathode materials has been focussed on improving upon

several well-defined structure types:

34

1. Pyrochlores such as R2MnRuO7 (where R = Tb, Dy, Ho, Er, Tm, Yb, Lu and Y),

feature good catalytic activity towards dissociation of oxygen combined with low

area specific resistance [82]. These have been investigated for use in both

cathodic and anodic conditions; for example, Gd2Ti1.4Mo0.6O7-δ has been shown

to be a sulphur-tolerant anode, by being able to operate on 10% H2S in H2

without suffering from significant degradation [83, 84]. However, these materials

generally contain rare earth or noble metals and so are hampered by high costs.

2. Layered double perovskites of the general formula AA'BB'O6, where A and A'

and B and B' are in alternate layers, such as (Gd/Pr)BaCo2O5+δ [85–87]. These

often feature high electrical and ionic conductivity at operating temperatures for

SOFC, however high activation energies reduce effectiveness at the lower end of

the operating temperature spectrum [88].

3. Ruddlesden-Popper type materials; An+1MnO3n+1 with A = alkaline or rare earth,

M = transition metal such as Lan+1NinO3n+1 (n = 1, 2 and 3) [89, 90]. Originally

reported for Srn+1TinO3n+1, these often contain a wide range of elements and

feature alternating blocks of perovskite and rock salt structures. They have

similar positive and negative characteristics to many novel perovskite structure

materials, namely high ionic and electronic conduction but with some chemical

reactions occurring between the cathode and electrolyte materials [89, 91].

4. Other perovskite types: there are several different perovskite structures which

appear promising, including variations around LaNi1-xFexO3 (LNF) and

Ba0.5Sr0.5Co0.8Fe0.2O3−δ (BSCF) amongst others [92–97]. Characteristics vary, but

often include high ionic and/or electronic conductivity at the expense of some

reactivity with common electrolyte materials.

35

Significant research has been put into cathode materials and while there are still

challenges to be overcome, great leaps forward have been made over the past few

decades with several materials well established for commercial use. This has not stopped

work being carried out into novel materials and of course the use of novel electrolytes

within cells, particularly IT-SOFCs, brings about a range of new issues such as chemical

compatibility and performance under different operating conditions.

2.3. Electrolyte Materials

Materials used as electrolytes in SOFCs must adhere to a range of characteristics,

including but not limited to: high degree of sintering to allow formation of gas-tight, thin

layers; high oxide ion conductivities; very low electronic conductivities; chemical and

thermal stability at operating temperatures in both reducing and oxidising conditions.

Currently, the majority of these characteristics are met by certain ceramics of the fluorite

or perovskite structure. The most common two materials in use within SOFCs are

stabilised zirconia and stabilised ceria. These are generally in the form of yttria-stabilised

zirconia (YSZ, fluorite-type structure) and gadolinia-doped ceria (GDC, also often

referred to as cerium-gadolinium oxide (CGO), fluorite-type structure). YSZ has high

oxide ion conductivity at temperatures over 800 °C, combined with excellent thermal and

chemical stabilities [98, 99]. It is also available at low cost and is already manufactured

on an industrial scale for use within oxygen sensors. However, the high temperature

required to reach adequate oxide ion conduction is a potential issue for SOFC durability

36

and has spurred research into alternative materials for use within intermediate

temperature SOFCs (IT-SOFCs). In the temperature region of 500-600 °C, GDC is

preferred as it has a similar conductivity to YSZ at 800 °C [30]. Above 600 °C in

reducing conditions, the Ce4+

is partially reduced to Ce3+

which reduces cell efficiencies

by allowing electronic conduction. Therefore the choice of conventional electrolyte

material is dependent on the temperature bracket for which the cell is designed. Ceria

may be doped with other rare earth metals to similar effect, as in samaria-doped ceria

(SDC).

A range of other perovskite materials for SOFC electrolyte use have also been

developed. One particularly interesting series is La1-xSrxGa0.8Mg0.2O3-δ (LSGM) where x

typically = 0.1 or 0.2, which has been extensively studied and used as a basis for many

other alternative systems [100]. LSGM displays a high oxide ion conductivity of ~0.14 S

cm-1

at 800 °C which is better than YSZ under the same conditions [101, 102]. However,

while this family of materials are generally chemically compatible with LSM, NiO in the

anode tends to react with LSGM during fabrication, forming LaNiO3 which can block

oxide ion transfer [103].

The wealth of research carried out into both perovskite (such as BaCeO3) and fluorite

(such as YSZ) structured materials for use as electrolytes in SOFCs has led to both

effectively reaching their maximum potential with little room for improvement [104],

and much current research is focused on optimising deposition techniques and

microstructural advances, in addition to investigations into the chemical compatibility

with novel electrode materials [105–112]. However, this has not prevented research into

37

other structure types in order to continue the pursuit of increased performance. Novel

structures investigated include:

1. Apatite-type materials: general formula of (Ln/A)10M6O26+y with Ln = rare earth,

A = alkaline earth, M = Si or Ge [113–115]. This family of materials typically

shows good oxide conductivity with negligible electronic conductivity. Further

research is being conducted with the aim of promoting the inclusion of Si over

Ge for cost reasons and to overcome potential chemical compatibility and

sintering issues [116].

2. La2Mo2O9: shows very high oxide conductivities above the phase-transition point

of ~580oC [117]. Work is continuing to synthesise a sample with high

conductivity at lower temperature, mainly through doping with the aim of

lowering the phase transition temperature [118].

38

2.4. Project Aims

The process of research into novel materials for use within SOFCs involves synthesis of

a large volume of materials, with subsequent dismissal of further research into certain

material groups. A variety of reasons may be the cause of abandoning research into

specific materials, such as impurities being present (from reagents not being fully

incorporated into the structure, for example), difficulties in sintering, poor electronic or

oxide ion conductivity and redox instabilities. Many of these issues are indicative of

greater problems such as poor mechanical strength or low performance as part of a fuel

cell.

The work within this thesis aims to synthesise and characterise novel materials, largely

focussed on materials with the perovskite structure. These are based around existing

systems, with the aim of improving stability and performance in reducing environments.

To this end, the synthesis and characterisation of novel perovskite materials was carried

out, with the goal of utilisation as an anode material within a solid oxide fuel cell.

39

Chapter 3. Experimental Techniques

In this Chapter, a description of the main experimental techniques used within this thesis

will be presented, to include the underlying theory that underpins their function and the

way in which they have been used to obtain data for this thesis.

3.1. Synthesis Techniques

This section covers a range of techniques suitable for use in synthesising metallic oxide

ceramics.

3.1.1. Solid State Synthesis

Solid state synthesis is one of the most widely used synthesis routes within materials

chemistry. It involves precursor materials, typically metal oxides/nitrates/carbonates

being combined in the correct stoichiometric quantities and then intimately mixed.

Combined powders may be ground by hand, or a mechanical mill (such as a vibromill,

planetary ball mill, etc) may be used to ensure more complete mixing. This mixture is

then placed in a crucible within a furnace and heated, typically to temperatures of 800 °C

or more, for periods of 12 hours or longer. This high temperature and long duration is

required to allow the new material to be formed through the breaking of the bonds within

the original materials and the migration of ions to new positions in the product. The

40

process of grinding and then reheating is repeated, usually with analysis by X-ray

diffraction between repeats to determine if the material has reached a single phase. While

simple to perform, the long reaction times and high temperatures required for the

diffusion of ions over a long distance can be disadvantageous, as can the lack of control

over microstructure and particle size.

3.1.2. Co-Precipitation

Co-precipitation involves creating a solution of metal cations with the correct elemental

composition for the desired material. This is typically achieved through using

stoichiometric quantities of metal nitrates dissolved to form an aqueous solution. A

precipitating agent is then added to the solution, causing a solid product to form. The

precipitating agents used vary widely but are often strong acids or alkalis. The solution

can then be filtered, leaving the solid precursor to the final product which is usually

formed through heat treatment. The temperature, concentration and pH of the solution

must be carefully controlled in order for the results to be as intended.

The advantages of this technique include the atomic level mixing of reagents which

affords a much faster route to phase purity than solid state synthesis, along with lower

temperatures required. However, it is more labour intensive and requires more equipment

than solid state synthesis which is an important consideration for shared laboratory

space. Initial material costs are also often higher which is particularly detrimental if

performing speculative synthesis. In addition, if the rates of precipitation of components

are different then a non-homogeneous mixture will result.

41

3.1.3. Sol-Gel

Sol-gel synthesis again involves the creation of a solution containing metal cations in the

correct stoichiometric quantities. Typically metal nitrates are added to deionised water

and heated to aid dissolution. Chemicals are then added to promote the formation of an

organo-metallic polymer; typically ethylene glycol and citric acid are used [119]. A

range of different ratios of ethylene glycol and citric acid have been reported as being

effective, with no definitive standard quantity universally used [120, 121]. The mixture is

then heated, evaporating water until a gel is formed. This gel is then heated in a furnace

to burn off the organic material, leaving a mixed metal oxide. The temperature required

for full organics removal typically range around 600 °C and care must be taken during

the heating process so that the remaining water does not evaporate too rapidly and cause

the mixture to overflow the reaction chamber.

Once again there are similarities to co-precipitation in terms of advantages and

disadvantages against solid state synthesis, including lower synthesis temperatures and

mixing of reagents at an atomic level, coupled against higher reagent costs and increased

labour intensity.

3.1.4. Hydrothermal synthesis

Hydrothermal synthesis involves utilisation of an autoclave (a pressurised reaction

vessel), typically stainless steel, which must be completely inert to the chemicals

42

involved in the reaction. The reaction mixture contains water and usually a strong acid to

aid dissolution of the reagents used. The reagents used are dissolved into the solution and

heated to the temperature required for the reaction to occur, above 100 °C to increase the

vapour pressure inside the vessel but typically under 250 °C. This combination of

supercritical water, high pressure and aggressive pH raise the solubility of many

materials, allowing the use of reagents which would normally not be water soluble.

Hydrothermal synthesis uses much lower temperatures than solid state synthesis, and

similarly to both co-precipitation and sol-gel synthesis the reagents are mixed at an

atomic level, which affords a much faster route to phase purity. The size and shape of the

products can also be controlled through varying the temperature, pressure and duration of

the reaction.

However, the reagents can be more expensive and, in some cases unavailable, making

this synthesis route suitable only in some cases. In addition it may be necessary to

optimise reaction conditions for every reaction to be carried out, with different

conditions leading to different product structures. It is also more equipment heavy,

requiring an autoclave and heater per reaction which are considerably more expensive

than the crucibles used in the solid state method. In addition, it cannot be used to

synthesise products that are unstable in the hydrothermal conditions used. Hydrothermal

synthesis is often utilised in growing single crystals of materials, and forming porous

structures such as zeolites which cannot be obtained by other means.

43

Within this thesis only solid state synthesis was used. Powders were weighed out using

an analytical balance accurate to ±0.1 mg. A planetary mill (Fritsch Pulverisette 7) was

used for grinding powders, with alumina crucibles used as reaction vessels.

3.2. Pellet Production for Conductivity and Dilatometry Experiments

Pellets of synthesised materials were formed through using a pellet press. This involves

using mechanical or hydraulic pressure to compress powder into a pellet within a die set.

The pressure required to form a pellet of sufficient mechanical strength to allow it to be

moved without damage depends on the properties of the materials used, including but not

limited to: particle size, moisture levels and material adhesion. In addition, binding

agents such as stearic acid, polyvinyl alcohol (PVA) and ethanol can be used to assist

pellet formation.

Once formed, pellets are placed into a furnace and heated to the desired temperature,

typically above 1000 °C. This has the effect of sintering together the particles which

form the pellet, increasing the density of the pellet and affording it much greater

mechanical strength.

Within this thesis, pellets were pressed using a manual hydraulic press, dies of between

8mm and 22mm diameter and pressures of up to 10 tons.

44

3.3. X-Ray Diffraction (XRD)

X-ray diffraction is a technique used for the analysis of crystalline structures which has

the capability to determine the dimensions of the unit cell and the crystal structure, and

so can give insights into the ordering of atoms within the structure.

X-rays are produced through bombardment of a metal target with electrons that have

been accelerated through a high voltage, typically around 40kV. This radiation is

powerful enough to eject core electrons from the metal, leaving a void in the core K (1s

orbital) shell. An electron from a higher energy shell, either the L (2p orbital) or M (3p

orbital), drops down to fill this void, and in doing so emits a photon of electromagnetic

radiation. As the hole which is filled is in the K-shell, this type of X-ray is called K-

radiation. The energy, and therefore the wavelength of the emitted electromagnetic ray,

is dependent on which shell the electron originates from; Kα is produced by an electron

from the L-shell and Kβ is a product of an electron from the M-shell. Spin multiplicity

causes these peaks to be split in two, forming the Kα1 and Kα2 peaks. In an attempt to use

only a single wavelength of radiation and so simplify the spectra obtained, filters are

used to block out the Kβ; in the case of the most common target metal copper, this filter

is made of nickel. In addition, monochromators can also be used, which can remove the

Kα2 peak, leading to a single wavelength, i.e. the Kα1 X-ray.

X-rays typically have a wavelength of around 1 x 10-10

m or 1 Å, which is in the same

region as interatomic distances in crystal structures. This similarity causes the X-rays to

be diffracted when passing through the solid, with the diffraction angle related to the

45

distance between the lattice planes involved which is quantified by Bragg’s law (see

Equation 6 and Figure 4).

(6)

Equation 6. Bragg's law, where n = integer, d = crystal plane spacing, λ = wavelength

of X-rays used and θ = angle of the X-rays in relation to the plane

Figure 4. Two crystal planes with spacing ‘d’ showing reflection of X-rays

In order for peaks to be measured, constructive interference must occur. The extra

distance travelled by the X-ray diffracted by the second plane in relation to the first plane

must be a whole number of wavelengths larger; i.e. (abc) = nλ where n = integer. Ideally

for XRD, a single crystal would be analysed. However, it is often not possible to create a

Incident X-Rays Diffracted X-Rays

46

single crystal of a material for study so powders are commonly be used. Powder XRD

was developed for this very reason, and though it is not as powerful as single crystal

XRD for structural determination, providing a starting model is available it can still be

used to determine a crystal structure.

There are 2 commonly-used instrument orientations used in powder diffraction, which

have different capabilities although the diffraction pattern obtained is analysed using the

same techniques. Typically, 0.5 g of powder sample is typically required in order to

record a diffraction pattern at a large enough intensity in reflection mode (Bragg-

Brentano). However, smaller quantities of samples may be analysed in transmission

mode (Debye-Scherrer). In order to get the data as uniform as possible, the sample is

usually spun; this also assists in trying to capture the maximum number of orientations.

The reflected X-rays occur at an angle of two theta (2θ) to the incident rays. With both

transmission and reflection geometries, the sample along with the X-ray beam and

detector must move relative to each other to preserve focussing and allow intensities at a

range of 2θ values to be recorded. The data recorded can then be compared to XRD

patterns stored in a crystallography database in order to assist in characterising the

material and determine whether there are any impurities.

For perovskite materials, particularly cubic perovskites, the XRD pattern recorded is

usually very simple for a phase pure material. The peaks are typically very clearly

defined with no shoulders which makes analysis straightforward.

47

All data within this thesis were acquired on either a Bruker D8 or a D5005 utilising a

copper target. The X-rays used for analysis are Cu Kα1 (λ = 1.5406 Å) for the D5005 and

Cu Kα (λ = 1.5418 Å) for the D8 diffractometer.

3.4. Crystal Symmetry

There are two different general types of solid; amorphous and crystalline. Amorphous

solids have no long range order, with atoms and ions arranged randomly throughout, in

comparison with crystalline solids which comprise of a regular lattice of atoms/ions

forming the structure. The smallest section of these crystalline materials that can be used

to replicate the entire lattice is referred to as the unit cell.

Although there are a wide range of potential arrangements of atoms or ions within crystal

structures, every single type of unit cell can be described as one of the 14 Bravais

lattices. Within this group there are 7 different crystal class types of varying dimensions.

Each of these has at least one of the following sub-types, depending on the arrangement

of the lattice: primitive (P), body-centred (I), face-centred (F) or base (C). The full

arrangement of all 14 Bravais lattices and the 7 crystal classes can be seen in

Figure 5 and Table 1.

48

Figure 5. The 14 Bravais lattice unit cell types [106]

49

Symmetry Side lengths Bond angles

cubic a = b = c α = β = γ = 90°

tetragonal a = b ≠ c α = β = γ = 90°

orthorhombic a ≠ b ≠ c α = β = γ = 90°

trigonal/rhombohedral a = b = c α = β = γ ≠ 90°

hexagonal a = b ≠ c α = β = 120° γ = 90°

monoclinic a ≠ b ≠ c α = γ = 90 °C β≠ 90°

triclinic a ≠ b ≠ c α ≠ β ≠ γ ≠ 90°

Table 1. Unit cell symmetries, side lengths and angles

Miller Indices are used to describe lattice planes in crystal structures. These are

determined using notation involving 3 integers, h, k and l, which represent directions

from the unit cell origin. These Miller Indices can be used to accurately describe a plane

within a unit cell, as can be seen in Figure 6.

The position of atoms/ions within a unit cell are denoted as (x,y,z). For example, within a

perovskite unit cell the A cation can be described using the value of (0.5,0.5,0.5) as it sits

within the centre of the unit cell, half way along each of the x, y and z axes.

50

Figure 6. Miller Indices for several lattice planes [106]

51

3.5. Point Groups and Space Groups

The symmetry within a unit cell can additionally be described through the use of point

groups and space groups. Point groups assign a single point to remain fixed in which the

cell is reflected or rotated around (either proper or improper rotation), referred to as an

operation. There are 32 different point groups, describing the rotation or reflection

symmetry of the crystal structure. Space groups expand on this and combine the point

groups with Bravais lattices and include screw axis and glide plane symmetry operations.

This provides a total of 230 space groups which describe all possible crystal symmetries.

3.6. Rietveld Refinement

The structural analysis of crystalline materials through XRD is effective only if there is a

suitable technique for analysing the results. Traditionally this meant that only high

symmetry samples could be studied in detail as otherwise multiple overlapping peaks

made the determination of peak intensities impossible. In addition, the shape of the peak

is important; the peak area (area underneath the peak, or integral intensity) gives a true

measurement of peak intensity, with the width relating to strain or disorder within the

crystal structure.

In an attempt to resolve this and allow more complex structures to be resolved, Hugo

Rietveld began using computers to analyse XRD data by calculating a model which fits

the observed data most closely [122]. This was achieved through using least square

refinements until the model and observed data correlated. This was achieved through

52

using individual data points, rather than peak intensities, and performing least square

refinements until the model most closely matched the observed data. A model provides

data, with the least square method altering the model by changing selected parameters in

order to reduce the sum of the squared residual differences between the model data and

the observed values. The model is then further optimised until there is a strong degree of

correlation between calculated and observed data. Rietveld refinement is now a widely

used method for structural refinement and can be applied to a wide range of materials.

Over the years, many different computer programs have been developed which make it

possible to carry out Rietveld refinements from diffraction data. Some of these are

relatively simple, open-source programs while others allow more complex data analysis.

Within this thesis the GSAS suite of software is used to identify structural characteristics

of materials [123, 124].

For Rietveld refinement to be carried out, some information about the system must be

known. Atom positions, cell parameters and space group are required, as well as the

wavelength of the X-rays used to acquire the data. The atomic positions and cell

parameters can then be refined during the analysis process in order to more accurately

match the observed pattern and so more accurately determine the structure. During the

analysis it is also possible to vary a range of other parameters to better fit the observed

data, such as peak shapes, atomic thermal displacement values, scale factors and atomic

occupancy.

53

Once all data has been input and the observed pattern is in place, it is possible for a

calculated pattern to be generated. Altering parameters alters the calculated pattern,

which can be compared with the observed data as in Figure 7. It is then possible to

display the difference between the two diffraction patterns which will help to identify

how good a fit the model is to the actual structure.

Figure 7. Diffraction pattern undergoing refinement using the GSAS suite

It should be noted that the requirement of a reasonably accurate starting structural model

means that it is not possible to use the Rietveld analysis technique on samples of

unknown structure.

54

3.7. Thermogravimetric Analysis coupled to Mass Spectrometry (TGA-

MS)

Thermogravimetric analysis (TGA) involves investigating the relationship of a change of

weight in a sample to the change in temperature. TGA can be coupled with Mass

Spectrometry (MS) which is used to analyse any exhaust gases from the sample and

determine their composition. This is combined with the data from the mass loss to work

out exactly what is lost from a sample, at what temperature and in what amount.

In TGA, a sample is loaded onto a precise balance within a furnace. The temperature is

increased in line with a profile defined by the operator, with the process being carried out

in either air, H2 or an inert gas. The TGA-MS equipment used within this thesis was

manufactured by Netzsch, with the TGA being model number STA 449 F1 and the MS

being QMS 403 C. This equipment is coupled, in this case, to a MS setup which analyses

the exhaust gases from the furnace in real time, allowing the gain or loss of weight to be

linked to the change in exhaust output. This leads to identification of chemical changes

which are tied to specific temperature changes.

Mass spectrometry (MS) involves the ionisation of a sample which is then analysed to

determine the mass to charge ratio. In the case of TGA-MS, the exhaust gas is analysed,

so the sample has already been vaporised. Any molecules are ionised before being

accelerated by an electric field. They are then analysed by a mass analyser to determine

mass to charge ratio and then pass onto a detector which analyses the quantity of each

ionic species present.

55

The combination of both TGA and MS into a single technique allows for the analysis of

exactly what is being released by a sample at different temperatures, and in what

quantities. This can lead to a better understanding of phase changes, chemical and

thermal stabilities of samples, and hence allow for changes to be made to improve

properties.

The weight, temperature and change in both must be monitored very carefully during the

analysis period. TGA-MS can be used to analyse a range of materials, from organic to

inorganic, complex to simple. In this thesis, the use will be focused on determining

oxygen content within ceramic powders. This was carried out in an inert gas flow or in

5%H2/N2 in order to reduce the oxygen content, and from the mass loss observed

determine the original oxygen content. This use of the TGA doesn't require analysis by

MS, so no mass spectrometry data will be presented.

3.8. Dilatometry

Dilatometry is the determination of the expansion of a material with regards to

temperature. The majority of materials expand when heated and by using a dilatometer

this expansion can be measured and so the thermal expansion coefficient (TEC) can be

calculated. There are there are several obvious examples of the use of dilatometry such

as mercury thermometers.

For research purposes one of the most common dilatometer configurations is to have a

highly sensitive pushrod encased within a furnace. As the furnace increases in

56

temperature the change in sample length is measured and from this the TEC can be

calculated, as in Equation 7 below [125]. Dilatometers may also be configured so as to

allow samples to be heated in atmospheres other than air, such as hydrogen or argon, in

order to observe different material TEC properties in different conditions.

(7)

Equation 7. Expression for calculation of linear thermal expansion coefficients, where a

= TEC, Li = initial length, Lf = final length, Ti = initial length and Tf = final length

Thermal expansion of materials occurs due to the increase in average bond lengths

between atoms within a structure at increasing temperatures. For materials of different

chemical structures, the overall expansion coefficient can be expressed as the sum of the

different components of the content of the structure, weighted for the angle of the bonds

in relation to their relative position within the crystal lattice. For example, the rock salt

structure has identical bonds all at 90° to each other so the increase in lattice size in any

one plane is equal to the increase in the bond in that direction, whereas the perovskite

structure has bonds at angles not equal to 90°of each other and so the increase in lattice

size in any one plane is a composite of the components in that directional plane.

Within SOFCs, the TEC of materials used is a very important consideration. Ideally, all

TEC values for all materials used should be similar for any temperatures the cell may

experience; in the case of SOFCs that range can be between 0 - 1000 °C. As the range is

so large, differences in TEC are magnified over the whole temperature band so large

57

differences cannot be tolerated by the cell. If TEC values for materials in contact with

one another are widely different then, upon heating, cracks may form or delamination

may occur, which would at the very least cause the cell to become less efficient and at

worse may cause the cell to cease functioning and become a hazard, especially if the

anode and cathode gases are allowed to mix. It is for this reason that, wherever possible,

TECs of materials must be analysed to confirm material compatibility.

Within this thesis, dilatometry measurements were carried out using a Netzsch 409C

dilatometer running on a profile from 25 to 1000 °C at 15 °C min-1

. Samples were

prepared by ball milling powders (Fritsch Pulverisette 7 Planetary Mill for 1 hour at 350

rpm) before being pressed into thick pellets of 10 mm diameter and approximately 10

mm length. These were sintered at temperatures relevant to the material being analysed

before being polished and having their length measured accurately using digital callipers

(accurate to ±0.01 mm).

3.9. Electrical Testing

A variety of electrical testing methods were employed to determine the electrical

properties of the materials and cells produced within this thesis, and are covered in detail

below.

58

3.9.1. Four Probe D. C. Conductivity

The electrical conductivity of a material is an important characteristic for any fuel cell

material. The electrolyte layer requires a material that has little to no electronic

conductivity, whereas electrode materials benefit from high electronic conductivity.

There are several methods for quantifying the conductivity of materials, and within this

thesis, 4 probe DC conductivity was used as it is relatively straightforward to carry out

and is suitable for analysis of electrode materials.

Analysis is carried out by attaching electrodes to a sintered dense pellet or strip in the

configuration seen in Figure 8. Current is passed across the outer electrodes (1 and 4),

with the difference in voltage between the inner two electrodes (2 and 3) measured. From

this it is possible to calculate the resistance, which can then be used to calculate the

sample conductivity. The equations involved can be seen in Equation 8.

Figure 8. Electrode configuration for 4-probe DC conductivity

59

(8)

Equation 8. Calculation for 4 probe DC conductivity (where σ = conductivity, L =

distance between electrodes 2 and 3, A = pellet cross-sectional area, I = current passed

between electrodes 1 and 4, V = voltage difference between electrodes 2 and 3)

4 probe DC conductivity is a useful technique for determining the conductivity of

samples with relatively low resistivities. As such, it is not suitable for analysis of

electrolyte materials with high resistivities; these would better be analysed using

techniques which allow more accurate determination of higher resistivities, such as AC

impedance spectroscopy. While there are other variations of 4 probe DC conductivity

(such as Van de Pauw), work within this thesis was carried out using the method outlined

above, as previous work on similar materials in both pellet and strip configurations had

shown little variation between different techniques and so the simpler technique was

utilised [126].

3.9.2. Electrochemical Impedance Spectroscopy (EIS)

Electrochemical impedance spectroscopy can be used to determine the behaviour of

electrode or electrolyte materials. This will be discussed within this thesis primarily as

AC impedance conductivity and also area specific resistance. The two techniques are

closely related, sharing many similar characteristics.

60

3.9.2.1. AC Impedance Spectroscopy

AC impedance spectroscopy is used within SOFC material testing to provide information

on oxide ion conductivity. It involves passing an alternating current voltage across a

sample and measuring the resulting current response. If the voltage passed across the

sample is sinusoidal, the current response will also be sinusoidal, and typically the input

and response will be out of phase. Expressions for a sinusoidal voltage and the phase

shift between input voltage and output current can be seen in Equation 9 and Equation

10.

(9)

Equation 9. Expression for a sinusoidal voltage, where e = observed voltage at time t, E

= input voltage and ω = frequency (in Hz) times 2π

(10)

Equation 10. Expression for the phase shift between input voltage and output current,

where i = current, I = the maximum amplitude of the current and φ = phase angle

between input and output

Most materials have some degree of resistance and some degree of capacitance. They

can be modelled as a series with the resistance and capacitance in series. The presence of

a capacitance effect leads to the phase shift of the output current. If the material acts as a

pure capacitor, the capacitive resistance can be shown as in Equation 11. Equally, if the

material acts as a pure resistor, the current can be calculated through Equation 12.

61

(11)

Equation 11. Expression for the capacitive resistance Xc for a pure capacitor, where Xc

= 1/ωC

(12)

Equation 12. Expression for the current of a pure resistor, where R = resistance

(13)

Equation 13. Expression for the voltage-current relationship for a pure capacitor, where

Ē = voltage amplitude, j = √-1 and Ī = current amplitude

For a pure capacitor, the output current would be perpendicular to the input voltage. This

means that if the current were to be plotted on one axis (horizontally) and the voltage

plotted on the other (vertically) then both real components could be described by the

complex notation shown in Equation 13.

(14)

Equation 14. Capacitance and resistance components in series

(15)

Equation 15. Expression defining impedance Z

As V = IR, Xc must contain a resistive element. The voltage Ē, when applied across a

capacitor and resistor in series, must equal the sum of the voltage drops of each

62

individual element, as seen in Equation 14. Through combining Equation 13 and

Equation 14, it is possible to define the impedance vector, Z, as can be seen in Equation

15. The impedance is formed of two parts; the real and imaginary parts, as seen in

Equation 16.

(16)

Equation 16. Expression showing the two components of the impedance, where ZRe =

real part (R) and ZIm = imaginary part (Xc)

Figure 9. Example of a Nyquist-plot for ionic conducting solids, demonstrating different

physical phenomena. Z’’ = imaginary impedance, Z’ = real impedance

This data obtained through EIS is often expressed in a Nyquist plot, where the imaginary

part of the impedance (Z'') is plotted against the real part of the impedance (Z') as a

function of frequency (ω). An example Nyquist plot can be seen in Figure 9.

63

Different physical phenomenon cause capacitance values of varying amounts, which can

often be easily identified on a Nyquist plot. For example, ionic resistance in the bulk of

the material would have a capacitance value in the region of 10-12

F, whereas that of the

grain boundaries within the material would have a capacitance value of 10-11

- 10-8

F.

The bulk and grain boundary resistances can manifest themselves as separate curves on

the Nyquist plot, however it is not uncommon for the curves to be partially or fully

merged, especially when the resistance of one is significantly greater than that of the

other. If impedance spectra are collected over a range of temperatures, it is possible to

calculate the activation energy as per Equation 17.

(17)

Equation 17. Expression for the calculation of activation energy as a function of the

properties of the sample and the temperature, where σ = conductivity, T = temperature,

A = sample surface area, Ea = activation energy and R = resistance

The activation energy can easily be found by plotting Log10(σT) / S cm-1

K vs T-1

/ K-1

,

with the activation energy being the slope of the resultant line.

The system analysed can be modelled as an electrical circuit, with the behaviours of

different components represented through capacitors and resistors. There are several

physical behaviours for various different cell components that have equivalent circuit

elements, which are listed below.

64

Polarisation resistance: also known as overpotentials, these occur whenever the

potential of an electrode is forced away from its OCV value. These can be split

into three separate components; activation polarisations, caused by the reaction

kinetics within the electrodes; ohmic polarisations, caused by the resistance to the

transfer of ions through the electrolyte (and, to a smaller degree, the electrodes);

and concentration polarisations, caused by limitations on the availability of

reactants and the ability of the system to replenish consumed reactants.

Double layer capacitance: The capacitance effect of forming a layer of ions on

the surface of a catalytically active substance, allowing the storage of a small

amount of electrical energy. This may also be modelled by a constant phase

element, which helps to define a capacitor which is not behaving in an ideal

manner (common in a real cell).

Charge transfer resistance: the resistance of the transfer of charge from the

reactants to the system, and vice versa.

Warburg impedance: caused by the diffusion of reactants. This is low at high

frequencies and increases linearly with reduction in frequency due to the

additional distance that reactants have to travel.

Typically, cell components are modelled as having several equivalent circuit elements

acting simultaneously, allowing accurate representation of the properties of the

component. The various elements can be modelled either in series or in parallel in order

to build as correct a portrayal of the system involved.

65

3.9.2.2. Area Specific Resistance (ASR)

The area specific resistance measurements made in this work use a variation on AC

impedance spectroscopy in order to acquire data. The ASR of electrode materials

contributes to the ohmic resistances of the overall cell, and as such it is important to

quantify and minimise these resistances where possible. Determining ASRs involves

passing an alternating current of different frequencies across a sample and measuring the

impedance. The components of a typical impedance are displayed in Equation 18.

(18)

Equation 18. Typical impedance (Z) where Z’ is the real part, j is the imaginary unit, ω

is the frequency and Z’’ is the imaginary part

Data collected using ASR are also commonly expressed using a Nyquist plot, containing

similar features as those formed through A.C. impedance conductivity. While ASR can

provide complex data when carried out on a full fuel cell, the complexity is minimised

when performed on a simplified cell. Cells used within this thesis comprise of a dense

sintered pellet of either CGO or YSZ electrolyte onto which a symmetrical layer of

electrode ink was applied to both sides and then heated to bond to the electrolyte. A layer

of platinum paste was then applied to both sides with a strip of platinum foil to allow

connection to an external circuit. This allows the cell to be analysed through AC

impedance spectroscopy. The dense electrolyte pellet, which has an electrical resistance

several orders of magnitude larger than that of electrode materials, appears on the

66

Nyquist plot separately so identification of the electrode and electrolyte response is

straightforward to carry out.

The area specific resistance can be calculated using data from the Nyquist plot. The

resistance forms a semicircle which crosses the x axis at least once. If a circle is fitted for

this data, the difference between the intercept points for the electrode response can be

used to calculate the ASR as in Equation 19. It should be noted that ASR analysis can be

carried out without fitting, simply through use of the intercept points on the x axis.

(19)

Equation 19. Calculation to determine the ASR from the intercept points on a Nyquist

plot, where IH = high intercept, IL = low intercept, n = number of sides (in the case of a

pellet, always 2) and A = area of one face

Within this thesis, ink suspensions were prepared containing 1:1 wt% of electrolyte and

electrode materials with 20%wt Decoflux (Zschimmer and Schwarz) to act as a

dispersant and binder. These inks were then ball milled (Fritsch Pulverisette 7 Planetary

Mill for 1 hour at 350 rpm) and coated onto dense pellets of Ce0.9Gd0.1O1.95 (CGO10,

Aldrich, presintered at 1500 °C for 12 hours) to form a symmetrical cell and then fired at

1000 °C for 1 hour. Pt paste was then applied to both sides to act as a current collector

and heated at 800 °C for 1 hour. ASR values were obtained in both cathodic (air) and

anodic (5%H2/N2) conditions at temperatures up to 800 °C (air) / 600 °C (5%H2/N2).

Impedance data were collected between 5 Hz - 13 MHz with an AC signal amplitude of

67

100 mV (HP 4912A Frequency Analyser), setting which had previously shown to give

reliable and repeatable results with few anomalies [126]. The spectra were plotted using

ZView software [127] to give an estimate of the resistance and capacitance of the

different cell component materials.

3.9.3. Fuel Cell Testing

Fuel cell testing was carried out on both microtubular cells and simple button cells. The

test rig was custom built using insulating fire blocks (Thermal Ceramics) and ceramic

rods. Nichrome ribbon (Advent Ni80/Cr20 1.5 mm x 0.13 mm) was used for heating

elements, wound around the ceramic rods. The blocks were hollowed to allow a heating

chamber to be formed. A Eurotherm 2404 controller was used to control the formed

furnace. Unit Instruments 7300 mass flow controllers (MFCs) regulated gas flow to the

system, via Swagelock piping. The tubes undergoing testing were inserted onto an inlet

pipe with an external diameter of approx 1.5 mm which allowed fuel to be supplied to

the anode layer on the inside wall of the mSOFC. Gases were supplied by BOC, with

99.95% pure H2 and 50:50 He/H2 being used. A schematic of the equipment used can be

seen in Figure 10.

68

Figure 10. Schematic diagram of the fuel cell test equipment used in this thesis

The button cells required additional equipment in order to make them gas tight. Cells

prepared were coated with porous silver ink paste (SPI) over the surface of both

electrolyte layers, and used to attach a short length of silver wire (Silver Wire Co,

99.98%) to either side. The button cell was then cemented onto a ceramic firing tube

with an outer diameter matching the size of cell, using ceramabond (Aremco) and left to

set for 6 hours. Once set, the same test procedure for the m-SOFCs was followed; the

firing tube with the button cell was placed over a length of Swagelock® tubing within the

furnace, allowing hydrogen to be supplied to the anode side sealed within the firing tube

and allowing air to reach the cathode side still exposed. A schematic showing the test

setup in more detail can be seen in Figure 11.

69

Figure 11. Schematic diagram of the pellet cell test setup used in this thesis

A Solartron 1400 was used to record all electrical measurements and to control electrical

demands on the cell. The Solartron CellTest suite of software was used to program a test

profile and to analyse results obtained.

3.9.3.1 Fuel Cell Thermodynamics

As a fuel cell is an energy conversion device, the efficiency of conversion is of

paramount importance. This can be calculated theoretically and also verified

experimentally. The energy available for conversion to electrical work is given by the

Gibbs free energy change for the overall cell reaction, which is the difference in the

energy of formation of the products and reactants and is the maximum amount of energy

70

that can be produced from the reaction under the conditions specified. The Gibbs free

energy change at STP for the reaction between hydrogen and oxygen to give liquid water

can be seen in Equation 20.

ΔG = GΦ

f , products - GΦ

f , reactants

kJmol-1

(20)

Equation 20. Gibbs free energy change of the reaction of hydrogen and oxygen to form

water

The Gibbs free energy change is itself a function of temperature and reactant

concentration (equated to partial pressure for gaseous reactants) and can be seen in

Equation 21, where is the standard Gibbs free energy change, R and T are the

universal gas constant (8.134 J K-1

mol-1

) and absolute temperature, and / / are

the partial pressures of H2O, H2 and O2 respectively.

(21)

Equation 21. Gibbs free energy change for the reaction of hydrogen and oxygen to form

water, with respect to partial pressure of gases

If the system is assumed to be operated reversibly, with all Gibbs free energy being

converted to electrical energy, then the electromotive force between the anode and

71

cathode takes the value of E in Equation 22, with n being the number of electrons

involved in the reaction and F being Faraday's constant (96485 C mol-1

).

(22)

Equation 22. Calculation of electromotive force between anode and cathode in a fully

reversible system

Combining Equation 21 and Equation 22 leads to Equation 23, the Nernst equation, in

which EΦ represents the reversible potential under standard conditions (1.23 V for a

hydrogen/oxygen system).

(23)

Equation 23. Nernst equation for the reaction of hydrogen and oxygen to form water

The Nernst equation relates the open circuit voltage (OCV, represented by E) of the fuel

cell to the temperatures an (in this case) partial pressures of reactants and products.

Typically for an SOFC running on hydrogen at 800 °C the open circuit potential is

between 0.9 - 1 V.

3.9.3.2 Fuel Cell Polarisation Curve

A fuel cell undergoing evaluation is usually subject to testing to determine the current-

voltage response of the cell. By recording the response of the cell to difference loads

72

placed upon it, it is possible to form a fuel cell polarisation curve, an example of which

can be seen in Figure 12.

Figure 12. Example fuel cell polarisation curve showing the three main sources of

voltage losses

The voltage losses within a fuel cell polarisation curve can be split into 3 main

categories: activation polarisation losses, ohmic losses and mass transport losses.

Activation polarisation losses are primarily caused by the kinetics of the reactions

taking place at the electrodes; for example, the oxidation of hydrogen to protons.

Ohmic losses are primarily caused by resistances in the materials used within the

cell; for example, the electronic resistance of the electrode materials.

73

Mass transport losses are primarily caused by the inability for enough reactants to

be supplied to the active layers of the electrodes to supply demand.

Developments in SOFC anode materials may assist in reducing losses caused in any of

these three sections. For example, improvements in electronic or oxide ion conductivity

would potentially reduce ohmic losses, while improvements in catalytic activity may

bring about a lowering of activation polarisation losses.

3.10. Microtube Manufacture

The processes used in the manufacturing of SOFC microtubes are detailed within this

section. A SOFC microtube (m-SOFC) is a solid oxide fuel cell with a tubular geometry,

typically in the region of 1 - 20 mm diameter. The tubular nature affords many

advantages, such as far easier gas sealing with respect to planar SOFCs, and small tube

diameters which allow relatively fast heating and cooling times. The m-SOFCs

comprises of several layers, with electrode materials on the inner and outer surfaces of

the cell and the electrolyte layer separating the two. Figure 13 displays an example of the

particular variety of anode-supported m-SOFC investigated within this thesis.

74

Figure 13. Schematic of an anode-supported m-SOFC [111]

3.10.1. Powder Preparation

Constituent powders were first milled using a vibromill to break down any particle

agglomerates and to ensure all solid materials had been mixed; this process is common in

cermet production [128]. This functions by rapidly shaking the sample within an

enclosed vessel containing milling beads, and can be used to break up particle

agglomerates, mix powders and reduce average particle size. It has also been noted that

ball milling of SOFC cermet powders can lead to increases in sample conductivities and

performance post sintering [129, 130]. This is believed to be due to the even distribution

of YSZ over the surface of the Ni particles and the interconnection between nickel

particles being increased, leading to a greater triple phase boundary area and subsequent

performance.

Powders were measured out in their respective quantities for the particular paste being

formed, and were then placed within a milling bottle. Ceramic milling beads were added

75

and the bottle sealed before being placed within the vibromill. The powders were milled

for 24 hours to ensure they were fully mixed and all agglomerates were broken down.

3.10.2. Paste Formation

The formation of a paste, which can be extruded to form a tube, is instrumental in the

overall process and the materials used can affect the performance characteristics of the

cell. The materials required for the anode are measured out in powder form and

combined with a binder which serves to hold the powders together. The binder is often

formed of both solid and liquid components, which when combined form an initial

amorphous solid [131]. The solvents used serve to soften the solid plastics to make them

more malleable and allow it to form a paste.

After being vibromilled, the liquid dispersants and binding agents were added to the

powders to begin paste formation. These were stirred by hand within a fume hood until

the powders began to form large agglomerates. The paste itself was formed using a twin

roll mill. This is a set of two smooth cooled metal cylinders, with a small adjustable gap

in between them, rotating in opposite directions at slightly different speeds. Photos of the

twin roll mill can be seen in Figure 14.

76

Figure 14. The Bridge twin roll mill used for this work

The mixture was added slowly to the two cylinders to form the paste while the distance

between them is gradually increased in line with the increased sample quantity. The

distance between the rollers was closely monitored and adjusted to cause shearing of the

paste to further aid mixing. The paste formed was tactile enough to adhere to the rollers,

77

typically more to the slower rotating cylinder than the faster. The rotation movement

forces it through the gap in between the cylinders which causes further mixing through

shearing. The desirable properties of the paste for this extrusion manufacturing technique

include it being soft and flexible enough to allow it to fold without cracking, and it

should also be possible to remove it from the cylinder without it sticking. It typically

took 10 - 15 minutes to achieve the correct consistency, during which time the materials

become well mixed. The paste was cut from the roller and re-applied to the cylinders

several times, in order to maximise mixing on the mill, before it was ready to use.

The twin roll mill utilised is of a suitable size to deal with batches of material as small as

50 g total of ceramic and metal powder, in addition to the binding formulation used. Any

smaller quantities would lead to the sample being too thin on the mill to work with easily

and would not provide enough material to allow extrusion to be carried out, while larger

batches can still be accommodated.

3.10.3. Cold Extrusion

Extrusion involves the compression of a material, which is then forced through an

aperture to cause it to form a specific shape. Extrusion is used to manufacture a wide

range of products, from steel girders to pasta [131]. There are many variations of

extrusion, however this thesis will focus on single barrel ram extrusion. This particular

type of extrusion involves a barrel into which the material to be extruded is placed. A die

set is installed at one end, with a ram inserted at the other. Pressure is applied via a

computer controlled load cell to the ram in order to force the material out of the die set,

78

see Figure 15.

Figure 15. Diagram detailing a basic extrusion technique

The extrusion die and equipment used here comprises of a barrel of 21 mm diameter onto

which a die is attached. This die is a metal cylinder with a central hole of 2.2 mm

diameter and a small indentation for an alignment cylinder. The extruder ram has been

manufactured with a thin cylinder of metal of 1.8 mm diameter (from here on referred to

as the mandrel) in the centre of the rod, which is designed to fit though the hole in the

die, causing the material present to be extruded as a tube with walls of 0.2 mm thickness.

The indentation in the die is to allow for a small brass cylinder, with the same diameter

hole as that of the mandrel, to fit onto the end, outside of the extrusion barrel, which

aligns the mandrel central to the die. This is forced out of the way once extrusion begins

and is designed to ensure that the walls of the tubes produced are as uniform as possible.

Co-extrusion was initially considered for this project, however was later ruled out. This

was due to several reasons: the required extensive work on paste rheology for pastes

79

comprising of novel materials would reduce the ability to quickly prepare tubes of a

variety of materials; the complexity of die sets increases the difficulty of re-engineering

if required; and that control of the thickness of the layers of material has proven difficult,

leading to difficulties in comparison between different materials produced through the

same technique. Overall, if the intention would be to produce tubes on a larger scale than

a prototype basis, co-extrusion would be well suited as it could considerably cut down

the production time of a batch of tubes, if properly controlled.

Photographs of the extrusion ram, mandrel and Instron extrusion equipment can be seen

in Figure 16, Figure 17 and Figure 18.

Figure 16. Extrusion rod complete with mandrel, without and with extrusion die and

alignment cylinder

80

The paste is prepared for extrusion by cutting to size and wrapping it tightly around the

mandrel to the width of the extrusion ram before inserting the extrusion ram into the

barrel. The correct size is to allow 5 mm at the end of the mandrel to be exposed which

can then be used to align the equipment, utilising the brass collar to confirm central

alignment. The die and barrel setup is then installed into a set of extrusion equipment, in

this case an Instron 4467 with a 30 kN load cell.

Figure 17. Extrusion ram and mandrel with YSZ/NiO paste wrapped around it. (L) shows

before insertion into extrusion barrel, (R) is setup complete with die as it would be inside

the barrel

A simple program is run to extrude the paste into a tube. The equipment is set to

compress at a rate between 0.5 mm and 2 mm per minute, which leads to the tube being

extruded at approximately 1 mm per second. On starting the extrusion process, the small

81

alignment piece is forced off the end of the mandrel by the extruded tube, which then

continues to be forced out of the die. These tubes are cut into lengths and laid on drying

racks to allow the solvents to evaporate, causing the binders to harden and give the tubes

temporary mechanical strength, before debinding and sintering of the ceramic takes

place.

Figure 18. Instron 4467 with 30kN load cell and sample installed (L), close-up of sample

ready to extrude (R)

Much of the powder processing, paste formation and extrusion work within this thesis

was based on methodology by Taylor et al. at the University of Birmingham [15].

82

3.10.4. Hot Extrusion

Hot extrusion was considered for use in this project. The process can be performed in a

variety of ways, all involving the use of a binder which is solid at room temperature and

liquid when heated. The heated medium is extruded through a die, as above, to form the

extrudate into the desired shape.

Work was carried out into use of a linear single screw hot extrusion machine for

extruding YSZ/NiO anode-supported microtubes. This equipment comprises of a hopper

containing the desired materials and binder at one end, a single archimedes screw which

forces the material through the system, a heating barrel which surrounds the screw to

provide a heat input, and a die set to form the extrudate to the shape required. Once

formed and cooled, these tubes can be sintered in the same fashion as those formed

through cold extrusion. Within this thesis, a Rondol Linear 18 single screw hot extruder

was used.

3.10.5. Debinding and Sintering

The debinding process removes the plastic and organic binding agents used to form the

paste. As this leaves the tubes very fragile it is combined with the first stage of sintering

the ceramic involved, which reduces handling and helps to minimise damage to the

tubes. This interconnection of ceramic gives the tube its mechanical strength and

provides the support for the other layers.

83

As debinding of organic materials usually occurs over 200 °C but sintering of most

ceramics requires at least 1000 °C (and often far higher), the temperature profile for the

overall process must be well structured in order to complete debinding and sintering

without the thermal stress of heating and cooling damaging the cells, as well as also

minimising the amount of time taken. The profiles used are detailed, along with various

binding and anode mixtures, within the extrusion experimental section of this thesis.

3.10.6. Layering

Each anode-supported microtube must be layered with both electrolyte and cathode

materials once the anode has been sintered. This can be achieved using a variety of

techniques, the most common of which are electrospray deposition, spray coating and

dip coating. The electrolyte layer should be as thin as possible, while still being a

complete and unbroken layer, in order to minimise resistance while preventing short

circuiting. The cathode layer should have high porosity in order to afford a large surface

area, while remaining complete and unbroken, so as to allow the easy collection of

current produced.

3.10.6.1. Electrospray Deposition

Electrospray deposition involves using an electric charge to produce an aerosol from a

liquid sample. The sample contains, either in solution or suspension, the materials which

will form the layer required. The liquid is electrically charged and then passed through a

jet nozzle which causes it to form many small droplets, kept apart by coulomb repulsion.

84

The size of these droplets can be tuned by changing the charge and the nozzle, amongst

other factors. This aerosol is deposited on a ground surface, losing its charge in the

process, and forming a layer. The technique is highly adjustable and can produce layers

of nearly any thickness, making it very well suited for electrolyte deposition.

Electrospray deposition of thin layers for use within SOFCs has been studied at length,

for example the deposition of YSZ for use in IT-SOFCs [132]. It is also possible to apply

layers of different materials over one another, potentially speeding up the fabrication

process.

However, the equipment is costly and time consuming to calibrate and prepare, which

could only be justified in this case if tubes were to be produced on a large scale. As work

was to be focused on small batches of tubes, this technique is less appropriate. If tube

production were to be scaled up or made commercial, electrospray deposition would be

amongst the recommended layering techniques, as the preparation time and cost could be

accounted for.

3.10.6.2. Spray Coating

Spray coating is similar to electrospray deposition in that it is the production of an

aerosol (containing materials of interest), which is deposited on a surface. However, it

usually uses compressed gas to both produce and propel the aerosol, rather than a

current, and this aerosol is directed at a target. It has the advantage of being relatively

quick to set up and the equipment required is cheap. However, it is much more difficult

to accurately determine and alter the thickness of the layer produced, making it less

85

suited to larger scale production. This inability to control the layer thickness to any

degree of accuracy also renders the process less repeatable. In addition, spray coating can

lead to a large wastage of material which would significantly lower the efficiency of the

process and the waste material would need to be recovered. Spray coating has been

successfully deployed to deposit SOFC materials such as LSCF, BSCF and LSM

cathodes, SDC protective layers and YSZ electrolyte layers [28, 133–135].

3.10.6.3. Dip Coating

Dip coating involves immersing a surface in a sample liquid containing either a solution

or suspension of the desired layer material. When removed from the suspension, the

liquid dries leaving a layer of solid behind. The process can be repeated in order to build

up thicker layers, if required, and can also be used to deposit layers of different

materials. The technique is made far more repeatable by utilising an automated machine

to control the speed of motion and duration of immersion, and has been carried out

successfully to form layers of common SOFC materials, such as electrolyte layers

comprising of YSZ [112]. The thickness of layers produced can be varied through

changing the formulation of slurry used. This technique falls somewhere between

electrospray deposition and spray coating with regard to ease and preparation time, as it

requires more expensive equipment and more time to set up than the latter, but less time

and less expensive equipment than the former.

86

3.10.6.4. Vapour Deposition

Vapour deposition allows a thin film layer of material to be deposited onto a target

surface. There are two main types of deposition; physical vapour deposition (PVD) and

chemical vapour deposition (CVD).

3.10.6.5. Physical Vapour Deposition

PVD is carried out by bombarding a source material with energy in order to vaporise it,

before this vapour is deposited onto a target surface. There are several different methods

which can be employed to vaporise the source, such as sputtering, electron beam,

cathodic arc and evaporative, all of which require specialist equipment and have various

advantages and disadvantages. They generally allow a specific layer thickness to be built

up and the microstructure of the layer produced can also be altered in some cases, such

as 1 μm layers of CGO barrier layers within SOFCs [136]. However it can take a long

time to set up in order to be able to reproduce and control the layers, and the equipment

can also be expensive which renders it unsuitable for this study. In addition, not all

materials are suitable for PVD application due to the high energies involved to induce

vaporisation, which can lead to disintegration or alteration of the source.

3.10.6.6. Chemical Vapour Deposition

CVD is widely used in the electronics industry as a large-scale method of applying a

controllable thin film of material onto surfaces. It involves the deposition of a layer

87

through a chemical process and so involves precursor materials which undergo reactions

to form the desired product material. The conditions for the process can be widely

varied, allowing liquid, gas and plasma based processes combined with variable vacuum

conditions to produce layers of precise thicknesses and microstructures. However, like

PVD, the expense and time taken to calibrate the equipment is a significant drawback

with relation to this project, and so CVD was not considered. CVD has been used to

deposit YSZ-containing cermet layers, amongst other materials [137].

Both spray coating and dip coating were used to prepare tubes documented within this

thesis, as equipment was available and could be tailored to the requirements of the

project in a fairly short amount of time. In addition, the smaller quantities of tubes

produced make both techniques highly suitable, with spray coating being favoured

overall due to the thin layers produced.

3.11. Scanning Electron Microscope (SEM)

SEM is a technique involving the use of a scanned beam of electrons to image a surface.

Due to the narrow nature of the electron beams used, SEM can produce high resolution

images of features down to approx. 1 nm in size, with a magnification of up to 500 000

times [138]. This is several orders of magnitude better than the most advanced light

microscopes, due to their use of electromagnetic waves of much longer wavelength than

the electrons used in SEM. In addition, back scattered electrons (BSEs) reflected from

the sample are used to differentiate areas of the sample containing elements of different

88

atomic number. This is made possible due to the heavier atoms interacting more with an

electron beam than lighter ones, so a larger proportion of electrons are reflected.

If it is desirable to determine the exact elemental composition of an area of a sample

examined, many SEMs have the ability to analyse samples using energy-dispersive X-

ray spectroscopy (EDX, also known as EDS or XEDS). When the electron beam

interacts with the sample it excites some of the electrons from the ground state in the

sample into an outer shell vacancy, creating an inner shell vacancy. When an electron

from a higher energy shell fills this inner shell vacancy it gives off a photon of energy in

the X-ray wavelength. Due to differences between the shell energy levels in all elements,

the X-rays given off are unique to each element and can be used to identify the species.

Utilising SEM and EDX in tandem can be particularly useful when studying the change

between aged and new SOFC cells as it can show species migration and agglomeration

of particles and can be used to identify the elements involved. In addition the effect of

sintering can be observed by analysing cells before and after. However, like many X-ray

based techniques it is difficult to analyse organic-containing components such as carbon

deposits within SOFC cells due to the light size of the elements involved, as X-rays

interact weakly with lighter elements.

A standard SEM setup usually involves an electron gun which is principally a cathode

made of tungsten. This is heated and charged using an electrical current which causes the

emission of electrons. Tungsten is often used due to its low cost and high melting

temperature, which allows it to remain a reliable electron source for a long working life.

89

The beam can be altered in energy by changing the temperature and the current passed

through the cathode, typically between 0.5 and 40 keV. This beam is focussed by lenses

to a typical diameter of between 0.5 and 5 nm. This flexibility on energy and beam size

allows different materials and structures to be imaged in high clarity. The interaction of

the beam with the interaction volume of the sample is measured through the detection of

the elastic and inelastic scattered electrons, and analysis of their energy and position

along with the detection and analysis of any electromagnetic radiation emitted. As the

technique scans over the surface it allows a picture to be built up, which is normally

stored as a digital image.

Most SEM analysis is conducted in a vacuum, due to electron beams being dispersed by

gases, which would lead to interference. In addition, samples must be electronically

conductive and ground at the surface in order to record an image of a sample. This

prevents any build-up of electric charge while the sample is being analysed. While this

makes metal items very straightforward to prepare for SEM imaging, non-conductive

specimens must be mounted on a metal sample stub before being coated with a

conductive material, typically gold. This is either performed using sputter coating or

evaporation; with the former being most common due to the low vacuum requirements

(evaporation needs a high vacuum). The layer formed allows all charge built up to be

dispersed through the stub. For the purposes of this thesis, all specimens analysed were

prepared on sample stubs, and coated with gold via sputter coating.

The equipment used within this thesis is a Philips XL-30 scanning electron microscope.

The information gained through SEM imaging in this thesis was used to analyse the

90

dispersion and microstructure of particles and materials within formed SOFC

microtubes, along with details such as the porosity of the material. In addition, the size of

particles or layers could be measured accurately and this information could be used to

identify potential ways to improve the production techniques of both materials used in

making the SOFCs, and also the procedures used to manufacture the SOFCs. For

example, specimens which display poor porosity in both anode and cathode layers could

be improved in several ways; either the manufacturing composition could be changed to

include more, or a different type of pore forming agent (PFA) in the mixture of materials,

or the materials themselves could be synthesised with different size or shape of particles

in order to attempt to improve the porosity.

91

Chapter 4. Extrusion and Fabrication of Microtubular

SOFCs

4.1. Introduction

Microtubular and planar SOFCs operate with similar principals; however, they have

various advantages and disadvantages over each other. Planar SOFCs typically have

higher power density and are simpler to manufacture; however, due to their size, they

typically suffer from slower start-up times which are necessary to ensure even heating of

cells. For these reasons, planar stacks of cells are more suited to continuous power

supply systems with few on-off cycles.

Microtubular SOFCs, however, are better suited to applications where faster start-ups are

required. Cells have been shown to be stable when heated at rates of up to 100 °C min-1

[139] and small stacks of cells have been shown to be relatively straightforward to

design [140]. Due to the geometry, they are also much easier to seal than planar cells;

however, there are usually more cells involved in a single stack so cell interconnections

become more complex. Smaller diameter cell sizes, while bringing about more

interconnections, also increase the power density of a stack, so some trade-off is required

between power density and complexity.

92

There are a variety of manufacturing techniques which are suitable for the production of

microtubular solid oxide fuel cells. Several stages are required for the production of cells

from dry powders. These stages can be broken down into four individual groups: Tube

Manufacture, Electrolyte Deposition, Cathode Deposition and Cell Test Preparation.

Each stage has several sub-stages for separate processes. The procedure followed within

this Chapter was based around work carried out by Taylor et al [15] and can be seen in

Figure 19.

Figure 19. Flow diagram for microtubular solid oxide fuel cell manufacture

Tube Manufacture

• Powder Processing

• Paste Formation

• Extrusion

Electrolyte Deposition

• Ink Preparation

• Dip Coating

• Sintering

Cathode Deposition

• Ink Preparation

• Dip Coating Ink 1

• Dip Coating Ink 2

• Sintering

Cell Test Preparation

• Silver Ink Application

• Silver Wire Application

• Cell Reduction

93

4.2. Experimental and Results

4.2.1. Paste Formation

The first stage of the project detailed in this thesis was to further refine existing

techniques so as to develop a manufacturing procedure which would afford reliable and

repeatable results with every batch of microtubes produced. The importance of this is

two-fold; it allows results to be checked and confirmed empirically, and it reduces waste

of some of the potentially expensive and hard-to-manufacture materials that will be used

later within the project.

Anode-supported tubes consisting of YSZ ceramic and either nickel oxide, copper oxide

or a mixture of both were chosen to be the first batches of tubes to be fabricated. This

choice of materials was made due to the fact that YSZ/NiO systems are well studied,

meaning that direct comparison with tubes purchased from an industry supplier could be

made, and the particle sizes should allow a framework of larger YSZ particles

surrounded by smaller NiO particles to form. Copper was chosen to be used as part of the

cermet in some batches of cells for several reasons: several studies have shown

performance improvements in anodes containing copper when running on methane; also

additionally to provide an idea of whether processing techniques could effectively be

transferred from a YSZ/NiO system to those containing other materials, with CuO

providing a suitable material as it can directly replace NiO as part of the processing

precursors, allowing composites containing both copper and nickel to be produced [27,

29, 129, 141]. Table 2 shows the ratios of powders used during initial fabrication. The

94

YSZ (99.99%) used was from Tosoh, with the NiO (99.9%) from Novamet (grade F) and

the CuO (99.995%) from Sigma Aldrich. The average particle size for NiO and CuO

were ≤1 µm and ≤5 µm respectively.

For the purposes of the first extrusion run, 100 g of each sample of cermet material was

prepared, to allow for experimentation during the extrusion process. Quantities of

powder used can be seen in Table 2.

Sample Reference A B C

YSZ (g) 50 50 50

NiO (g) 50 - 25

CuO (g) - 50 25

Table 2. Quantities of powders used to make up the cermet for different samples

The powders were mixed together within a fume hood until a mixture of uniform colour

was obtained before being ball milled in a vibromill for 24 hours, along with 10 alumina

milling beads. Vibromilling was chosen as the mixing method for the dry materials as it

was shown to have a positive effect on microstructure and mechanical strength of the

produced tubes [142]. The solid binder materials were added and mixed using a mortar

and pestle before the liquid binders were added. The first batch of powders had binders

added in the quantities listed in the prior work [15]. This was found to yield pastes too

stiff to easily work with, and so several further batches were prepared using

proportionally increased quantities of binding materials up to 150%. It was found that

95

125% of the literature values was most easy to work with, values shown in Table 3.

Material Quantity per 100 g cermet

Cyclohexanone (VWR International) 10.5 g

Polyvinyl Butaryl (PVB) (Acros Organics) 9.5 g

Dibutyl Phthalate (Sigma Aldrich) 15 drops

Stearic Acid (BDH Chemicals) 0.125 g

Table 3. Binder materials used for paste formation

Binders based on organic solvents, rather than water, were selected for use within this

study. This was due to both previous work carried out at the University of Birmingham

and also due to the difficulties in producing reliable, straight tubes using water-based

binding agents [15, 143].

It is thought that the decreased particle size of the powders used in this thesis against

those used by Taylor et al (1 μm/5 μm vs 10 μm) accounts for why the different

quantities of binders were required, as smaller average particle sizes lead to

proportionally larger surface areas [15].

4.2.2. Extrusion

The three samples from the first successful paste batch, 1A, 1B and 1C, were extruded as

96

described in Chapter 3 to form tubes. However it could clearly be seen that, even when

set up as accurately as possible, the equipment used still failed to produce straight tubes

with regular wall thicknesses. The die set was inspected and found to be damaged and in

need of re-machining. While this equipment was being re-made, the next steps in

microtubular SOFC production were carried out on the initial 3 samples as a test of

principal, as well as to observe if it was possible to obtain a working fuel cell using the

technique.

Tubes were first laid out within straight, smooth channels on wooden drying racks to

allow them to harden as straight as possible. After 24 hours they were then fired in a

furnace within cylindrical firing tubes, with the firing profile dependant on the cermet

materials used. Firing within ceramic firing tubes was found to be necessary to minimise

bending of the microtubes upon heating. As CuO would begin to melt out of the tubes at

1083 °C, samples 1B and 1C, containing CuO, were fired at temperatures up to 1200 °C

in an attempt to find a balance between keeping the material present and sintering of the

YSZ. The reluctance of YSZ to sinter at this lower temperature meant that dwell times

needed to be increased in an attempt to prolong the sintering stage. In addition, a variety

of final dwell temperatures were investigated to ascertain the difference, if any, on the

final product. These profiles are listed in Table 4 and Table 5.

97

Process Start Temperature ( °C) End Temperature ( °C) Time (mins)

RT 150 120

Drying 150 150 60

150 600 450

Debinding 600 600 240

600 1300/1400 480

Sintering 1300/1400 1300/1400 720

1300/1400 ambient 180

Table 4. Firing Profile for sample 1A

The heating profile used is a slightly modified version to that found in the literature [15].

It was developed complete with slow rises in temperature of around 1 °C per minute in

order to minimise cracking of tubes. The profile incorporates drying the microtubes

before debinding and then finally sintering, allowing the tubes to stay in position through

all steps, in order to minimise mechanical stress before sintering has occurred. Heating

rates were kept relatively low in order to minimise thermal expansion of materials

causing cracking or delamination.

98

Process Start Temperature ( °C) End Temperature ( °C) Time (mins)

ambient 150 120

Drying 150 150 60

150 600 450

Debinding 600 600 240

600 1100/1200 480

Sintering 1100/1200 1100/1200 1080

1100/1200 ambient 180

Table 5. Firing Profile for samples 1B and 1C

Figure 20. Graph showing sintering profiles for NiO-only samples (batch 1A) and CuO-

containing samples (batches 1B and 1C)

0

200

400

600

800

1000

1200

1400

1600

0 500 1000 1500 2000 2500

Tem

per

atu

re (

°C

)

Time (mins)

NiO samples

CuO samples

99

Once fired, tubes were inspected for visible cracks or defects caused by heating, and also

for melting of the CuO present in samples 1B and 1C. Unfortunately, these copper-

containing samples were deemed to have been degraded through the loss of CuO even

with the lowered sintering temperatures used, as a significant amount of material had

melted onto the firing tubes used within the furnace. In addition the 1100 °C batches

remained relatively weak and were prone to snapping easily while being handled. The

1200 °C batches were also difficult to remove from the firing tubes due to increased loss

of CuO, which was found to be melting out of the microtubes and causing them to react

with the surface of the firing tubes.

The next stage was to deposit an electrolyte layer. Dip-coating was chosen out of the

deposition techniques as this was being developed in another project at the University of

Birmingham for use on SOFC microtubes bought from a leading microtubular SOFC

manufacturer, which were complete apart from the cathode layer, and this study helped

to further the technique and ink slurry development [144]. The initial YSZ electrolyte

slurry used is listed in Table 6.

Material Quantity

YSZ (Tosoh) 5 g

Acetone (Sigma Aldrich) 5.5 g

KD2 (Sigma Aldrich) 0.2 g

Terpineol (Sigma Aldrich) 1 mL

Table 6. Initial composition of electrolyte slurry used for dip coating

100

This slurry, along with 10 alumina milling balls, was vibromilled for 24 hours in order to

fully mix the materials, and also to reduce the particle size and particle conglomerates.

The tubes were prepared for dip coating by sealing and covering a small section of each

end in order to leave a section of anode exposed to allow an electrical connection to be

made. The YSZ slurry was then dip coated (at a rate of 0.5 mm s-1

) onto several tubes

from sample 1A, from both the 1300 and 1400 °C sintering groups. A sample of several

tubes were dip coated twice in an attempt to determine if the thickness of the electrolyte

layer formed had a noticeable effect on the fuel cell performance. Once dipped, these

tubes were left for 24 hours to dry. They were then sintered at 1300 or 1400 °C

(dependant on their original sintering temperature) for 12 hours.

Second and third slurries were formulated using quantities of YSZ increased by

50%/100%, i.e. 7.5 g and 10 g, while retaining the original quantities of dispersants and

solvents. These were formed in the same way, through vibromilling for 24 hours;

however, they proved to be too thick to be used easily and so were discarded. These

slurries were trialled in order to determine that the maximum level of YSZ was being

included so as to deposit an electrolyte layer in a single process, without requiring a

second dip. As the initial goal was to achieve a working cell, rather than an optimised

cell, having an electrolyte layer thicker than ideal would not be considered detrimental at

this early stage.

Cathode layers were then dip coated onto the microtubes. The microtubes that are

purchased from a leading microtubular SOFC manufacturer by the Fuel Cell Group at the

University of Birmingham consist of anode-support and electrolyte layers and require the

101

deposition of cathode material. This has led to a standard deposition formula being

developed [19, 20, 145]. There are 2 inks used for cathode forming, one which bonds

strongly to the YSZ-based electrolyte (cathode ink 1), and the second which then forms a

more active layer (cathode ink 2). The formulations can be seen in Table 7.

Material Cathode Ink 1 Cathode Ink 2

La0.5Sr0.5MnO3 (Praxair) 6.5 g -

La0.82Sr0.18MnO3 (Praxair) - 20 g

YSZ (Tosoh) 6.5 g -

Acetone (Sigma Aldrich) 0.25 g 0.4 g

Acetone (Sigma Aldrich) 16 mL 14 mL

Triolein (Sigma Aldrich) 0.2 g -

Terpineol (Sigma Aldrich) - 2 mL

Milling Beads 12 7

Table 7. Composition of cathode inks 1 and 2

These components were measured out before being vibromilled along with 10 alumina

milling beads for 24 hours. Both the 1300 °C and 1400 °C NiO/YSZ sintering tubes were

coated in the same way. Initially, samples were dip coated in cathode ink 1 at a rate of

0.5 mm s-1

, then left to dry for 24 hours. Cathode ink 2 was then deposited in the same

manner, followed by 24 hours drying time. Both sets of samples were fired at 1200 °C

for 12 hours to allow the binding of the cathode layers.

102

4.2.3. Re-Machined Die Set Samples

Once the die set had been re-machined it was used to fabricate several additional batches

of NiO/YSZ tubes. These were formulated in an identical manner to those in batch 1A,

including firing at both 1300 °C and 1400 °C, layering with both electrolyte and

electrode materials and preparation for cell testing by application of silver ink. These

batches will be referred to as batch 2.

The cells produced in batch 2 were more regular than those produced in either batch 1A,

1B or 1C. This regularity was noted through the fact that the tubes extruded using the re-

machined die set could be placed within the drying racks with much more ease, with

each groove in the rack being ~0.5 mm wider than the width of extruded tubes,

indicating a straighter extrusion. However, closer inspection revealed that wall

thicknesses still varied, and producing cells with walls of uniform thickness proved very

difficult. Many repetitions of the extrusion process were carried out as part of batch 2 but

despite these many attempts, including minor variations to the paste formulations and

alterations to the extrusion rates, reliable manufacturing of cells using static variables

proved impossible.

4.2.4. Hot Extrusion Samples

Several binders were trialled, recommended by the manufacturer of the extrusion

equipment and chosen for their different properties upon heating.

Polymethylmethacrylate (PMMA) and high density polyethylene (HDPE) were selected

103

as they both had melting temperatures within the operating window of the equipment

used (< 300 °C), were straightforward to source and provided a variation in crystallinity

(i.e. they had different softening behaviour on heating, different tensile strengths etc).

Mixtures of YSZ and NiO (50:50 by weight) were combined with various proportions of

binder, from 50 TO 100 %wt. These were then fed though the Rondol Linear 18 extruder

at varying temperatures to ascertain the optimum for tube production.

The ideal quantity of binder for production of even, uniform microtubes was found to be

in the region of 50:50 wt of binder to YSZ/NiO mixture for both PMMA and PEHD.

PEHD was found easier to work with, as the sharper melting point (i.e. the material

soften less up to melting point) allowed the temperature to be set just above the melting

point of ~130 °C, so that the tubes produced hardened faster on exposure to air.

However, on sintering using the profile listed above (for bath 1A tubes), all tubes

produced using both binders were found to have disintegrated. This was postulated as

being due to the relatively low proportions of cermet contained within the pre-sintered

tubes, approximately 20% by volume, which was insufficient to provide mechanical

strength.

Efforts were made to increase the quantity of cermet present in the extruded tubes, while

retaining uniform shape. 50 %wt for both binders trialled was found to approach the limit

of the equipment; below this threshold, there was too little binder present to allow the

paste formed under heating to flow through the die set without exceeding the torque

safety limits of the linear screw, even with extrusion temperatures increased in an

attempt to confirm that all binder was indeed molten. When using binder quantities

104

significantly below 50 %wt, the equipment used would seize regardless of the

temperature of operation due to having too little binder present to allow the flow of

paste. However, even with this reduced binder concentration, the produced tubes

disintegrated on sintering. While previous studies have shown that formation of a thin-

walled microtube consisting of YSZ8 was possible using approximately 50:50 wt ratios

of binder to ceramic, incorporation of metal oxide powders (such as NiO) weakens the

ceramic framework leading to tubes of insufficient mechanical strength [106]. In

comparison to the cold extrusion, which allows binders to be present at ~20 %wt

(approximately 1:1 by volume) or below, the minimum binder concentration of 50 %wt

for hot extrusion using the available equipment was a significant disadvantage and led to

the continuation of studies into microtube production by this method to be discontinued.

4.2.5. Cell Testing

NiO/YSZ tubes from both batch 1A and batch 2 were prepared for performance testing.

This involved applying porous silver ink paste (SPI) and silver wire (Silver Wire

Company, 99.98%) to act as current collectors, and connecting them to both the anode

and cathode. The cell preparation and current connection techniques used were based

around those developed by Dhir et al [19, 20]. The size of active area for each cell was

calculated by measuring the cathode: the cells produced were typically 1.6 mm in

diameter and the active area was 2.5 cm long, giving an active cathode area of

approximately 1.0 cm2.

Both 1300 °C and 1400 °C sintering tubes from batches 1A and 2 were tested as

105

complete fuel cells. This involved placing them in a firing tube in a hydrogen/helium

atmosphere at room temperature, before raising to 700 °C with a flow of 20 mL min-1

of

H2/He and holding for 1 hour to reduce the NiO to nickel metal, followed by testing to

obtain information on their performance. Reduction at this temperature and in this

fashion has been proven to be an optimum balance between mechanical strength and

performance on methane [19, 145]. Once reduced and cooled, microtubes were slotted

onto an injection tube which acted to hold the cell in position and also provide fuel to the

anode. This injection tube was made from stainless steel and the fuel was provided at a

variety of rates between 10 - 50 mL min-1

. Cells were heated to 800 °C at a rate of 20 °C

min-1

, mirroring testing procedure for the commercial cells. The initial measurements on

fuel cell testing, involving determining the open circuit voltage (OCV) and collecting

information to build a current-voltage curve (I-V curve) were carried out. The OCV is

the voltage recorded with a negligible current drain applied to the cell, with the I-V curve

being found through applying different current loads to the cell and recording the

voltages achieved.

Both sample groups of cells were able to record an OCV, with little noticeable difference

between the two sintering temperatures or batches, most initially recording a value in the

region of 1.0 V at 800 °C running on dry hydrogen as a fuel, with a few anomalous

exceptions. This indicated a cell that was sealed, with no fuel leaking, as expected with

the fabrication techniques used [140].

The testing profile for recording a polarisation curve is to draw incrementally larger

currents from the cell and record the output voltage, allowing the curve to be built up

106

from the data. As the cells were all in the region of 1 cm2 active area, and previous

studies have shown that commercial cells of similar materials can record current

densities in the region of 0.7 A cm-2

at 0.5 V, a test profile involving stepping the current

up in increments of 0.05 A from OCV was decided upon, with the test being halted once

the voltage reached 0.4 V. However, neither set of samples was stable enough to

withstand the current demands placed upon them during the I-V measurement stage, with

output dropping rapidly to the trip voltage. Examples of voltages obtained during this

process can be seen in Figure 21,with all cells being from batch 1A in this case.


As it can be seen, cells recorded show a very rapid decay in voltage as soon as there was

a current demand, with the test tripping at the low voltage mark within several seconds.

The cells have been offset by 1 second in order to separate the data better, with the

current demand being placed on at 7, 6 and 5 seconds respectively. The current demand

0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10

Vo

lta

ge

V)

Time (s)

Cell 1

Cell 2

Cell 3

107

was only 0.05 A at this point. Due to the fast voltage degradation, analysis into the cause

of the failure of cells was made more difficult; too little data was being recorded to

determine what exactly the failure mode was, or to generate a polarisation curve. In

addition, once this initial testing had been performed, attempts were made to record

OCVs once again, however readings were typically only a few mV, with no cells

reaching an OCV above 0.1 V.

The testing profile was then altered to record readings 5 times a second, up from 2 times

per second. In addition, the current demand was changed so that it increased in

increments of 0.01 A every 10 seconds until the low voltage trip of 0.4 V was reached.

Examples of voltages recorded against time through this process for cells from batch 1A

can be seen in Figure 22.


0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10 12

Vo

lta

ge

(V)

Time (s)

Cell 1

Cell 2

Cell 3

108

The voltages decay at similar rates for both test profiles, and within 2 seconds of drawing

a current the low voltage trip had been reached. Again, analysis of the failure mechanism

proved difficult as so little data could be collected before irrecoverable damage to the

cells occurred.

An attempt was made to obtain a polarisation curve using voltage control, reading the

current produced at different voltages and using this information to create a curve.

Again, data were recorded at 5 Hz, with the test profile starting at OCV and changing the

voltage demand by 25 mV every 5 seconds from 0.9 V down to 0.45 V. Again, data from

cells from batch 1A are shown here.

Figure 23. Voltages and currents for cells moving from OCV to 0.9 V voltage demand

As it can be seen in Figure 23, again within several seconds of testing the cell potential

drops and hits the trip point. This being despite drawing a specific voltage instead of a

current, attempting to allow the cell to reach a power which allows the cell to remain

0

0.2

0.4

0.6

0.8

1

1.2

0 0.005 0.01 0.015 0.02

Vo

lta

ge

(V)

Current density (A cm-2)

Cell 1

Cell 2

Cell 3

109

stable. It should be noted that, under all settings used here, the measurement error is 10

mV / 5 mA over the range of readings, meaning that the results shown above in Figure

23 have potentially large errors.

This was initially put down to the unreliable and damaged extrusion equipment not

producing regular tubes, which could therefore not withstand the thermomechanical

stresses which occur at the operating temperature and conditions for a SOFC. After the

tests, the tubes were carefully inspected. There was no visible damage to any cells, with

no obvious cracks or leaks, although it was apparent that cells were considerably more

brittle and fragile after exposure to fuel cell conditions than before. This was shown by

several cells breaking during handling and disconnection from the test rig, something

which did not occur at all during preparation for cell testing.

Analysis by SEM of cells at a variety of production stages was carried out in order to

view the microstructure of materials and determine whether any obvious weaknesses

could be identified which may be contributing to the poor stability displayed in cell

testing.

4.2.6. SEM Imaging

Tubes from batches 1A, 1B and 1C along with batch 2 (those fabricated using the re-

machined die set) were analysed using a scanning electron microscope, as detailed in

Chapter 3. Images obtained can be seen in Figure 24.

110

Figure 24. SEM images of tubes extruded of various batches and in various states of

production. a = YSZ-NiO-CuO as extruded, b = YSZ-CuO debound and sintered at 1100

°C, c = YSZ-NiO debound and sintered at 1400 °C, d = YSZ-CuO as extruded, e = YSZ-

111

NiO debound and sintered at 1300 °C then YSZ electrolyte deposited and sintered at

1300 °C, f = YSZ-NiO debound and sintered at 1100 °C, g = YSZ-NiO as extruded

As it can be seen in Figure 24a, before the thermal removal of the binder the anode layer,

which the tube is comprised of, is almost completely non-porous. The wall thickness

here can be seen to be approx. 200 µm. There is little noticeable difference across the

thickness of the cell wall, with few large agglomerate particles over 10 μm diameter.

Figure 24b shows a close up image of a sample B tube wall after it has undergone the

thermal removal of the binder and sintering at 1100 °C. It can be clearly seen that the

porosity has increased over the sample seen in Figure 24a, before debinding and

sintering. The loss of the binding agents used accounts for the increased porosity.

Figure 24c shows the decrease in porosity in samples sintered at higher temperatures,

due to increased YSZ sintering and NiO agglomeration. A noticeable difference can be

seen between Figure 24b and Figure 24c, showing the decrease in sample porosity due to

the increased sintering temperatures used. This additional sintering has partially

counteracted the pore-forming effect of removing the binding agents. The wall thickness

can be seen here to be approx. 150 µm.

Figure 24d shows the full profile of a tube manufactured from a batch of sample B paste.

As with Figure 24a it can be seen that the walls are non porous; however, it is clear that

the thickness of the wall varies quite dramatically at different points around the tube.

This range is between 75 - 200 µm. In addition, the tube is not perfectly round, even

before undergoing thermal binder removal and sintering, which may lead to further

112

deformation as in Figure 24g.

Figure 24e shows the tube wall of a cell after coating with a YSZ electrolyte layer. The

thin electrolyte layer can be seen on the outside of the cell wall and is easily

distinguished from the more porous anode layer. The overall wall thickness shown here

is in the range of 100 - 150 µm, with the electrolyte layer in the region of 10μm.

Figure 24f shows the semi-porous wall structure as caused by the thermal removal of the

binder. The difference between Figure 24d and Figure 24f is noticeable, with a

significant increase in porosity. The wall thickness here can be seen to be in the range of

150 - 200 µm.

However, Figure 24g shows the unreliable nature of the extruded tubes. As can be seen,

the walls are not of a uniform thickness, nor are the tubes round. In addition, the tube

wall is at one point very close to splitting completely, with a thickness of approximately

25 µm. This deformation was a common occurrence and through many extrusion runs

and SEM sessions it proved difficult to identify a sample which displayed a round,

uniform profile. However, it should be noted that deformities such as that present on the

upper left side of the tube pictured in Figure 24g may be due to the cell having been

snapped in the preparation process for SEM analysis.

The lack of reliability of formed tubes has implications for full fuel cell testing. With

tubes having irregular wall thicknesses and shapes, thermal stability of samples may be

seriously compromised. In addition, achieving repeatable results is made considerably

113

more unlikely without a repeatable method of tube production.

4.2.7. Further Extrusion Work

Once the die and ram set had been re-made, further extrusion work was carried out.

Multiple batches of YSZ/NiO paste were formulated as per above, however it proved

very difficult to extrude tubes of uniform wall thickness. It was noticeable that the re-

machined die set was easier to align, however once the alignment cylinder had been

displaced by extrudate the ram became misaligned almost immediately. Many procedural

modifications were made with the aim of improving tube shape. Paste batches were

formulated which had lower concentrations of binder materials, consequently making

them stiffer in an attempt to prevent the mandrel from becoming misaligned. Rates of

compression for the load cell were varied between 0.5 - 10 mm min-1

with a variety of

paste binder concentrations, also in an attempt to prevent the mandrel from becoming

misaligned and to try to increase the length of tube extruded before misalignment

occurred. However it still proved impossible to form repeatable, reliable tubes of

uniform profile.

Nevertheless, samples of extruded tubes were prepared for fuel cell testing using the

methods detailed above. However, these suffered from the same rapid degradation even

without a load on the cell, with OCV recorded dropping after a maximum of 15 minutes

to under 100 mV. Consequently, it was decided to vary the electrolyte and cathode layer

deposition techniques in order to attempt to counteract this failure and to verify that

thermomechanical stresses were the more likely cause of the problem, rather than layer

114

formation. As tubes purchased from commercial sources came pre-prepared with an

electrolyte layer, the cathode deposition technique was known to be successful through

painting the ink onto the cell by hand and so this was to be repeated on this third batch of

cells.

To this end, electrolyte layers were deposited using spray deposition. The ink utilised for

electrolyte dip coating (see Table 6) was found to be suitable for use with the equipment

available, being thin enough to avoid blocking the spray nozzle. Tubes (which had

undergone debinding and sintering at 1400 °C) were again sealed at both ends and

attached to a custom-made rig which span the sample at a uniform rate in an attempt to

deposit ink over the cell at a constant rate. Batches of samples were prepared, which had

one, two and three separate coatings of electrolyte slurry, and which were left to dry for

1 hour between applications. These were sintered at 1400 °C for 12 hours before having

cathode inks applied via the dip-coating technique and being treated in the normal in-

house fashion for m-SOFCs within the research group. Cells were prepared for fuel cell

testing as described above, with silver ink and silver wire used for current collection

purposes. Tubes were pre-reduced at 700 °C in H2/He for 1 hour before fuel cell testing.

It was found that these cells also underwent rapid degradation in the same way as the

dip-coated electrolyte cells, with OCVs initially in the region of 1.0 V, but dropping to

negligible values after attempting I-V measurement testing and also when running at

OCV for longer than 10 - 15 minutes. This furthered the speculation that the electrolyte

layer deposition technique was not at fault for the rapid performance drop, as several

different techniques had led to the same result. Examples of attempts to draw power from

115

cells can be seen in Figure 25, where a voltage demand of 0.9 V was placed on the cells.

Once again, data was recorded at 5 Hz to try to obtain a better insight into the cell

behaviour.

Figure 25. Voltages and currents for cells (extruded using the remanufactured die set)

moving from OCV to 0.9 V voltage demand

Samples were also prepared which had the electrolyte layer applied by depositing an ink

by hand. The ink formulation used in Table 6 was found to be too thin to easily apply

using a paintbrush so a formulation with 50% increased YSZ content (over the spray

deposition ink) was produced using the same process described above. The composition

of this can be seen in Table 8.

0

0.2

0.4

0.6

0.8

1

1.2

0 0.005 0.01 0.015 0.02 0.025 0.03

Vo

lta

ge

(V)


Cell 1

Cell 2

Cell 3

116

Material Quantity

YSZ 7.5 g

Acetone 5.5 g

KD2 0.2 g

Terpineol 1 mL

Table 8. Composition of electrolyte slurry used for manual deposition

This thicker electrolyte slurry was applied to YSZ/NiO tubes that had undergone

debinding and sintering at 1400 °C. Several different batches were produced, some of

which had just a single layer of electrolyte ink deposited and others which had two or

three layers applied with an hour drying time between applications. These were sintered

at 1400 °C for 12 hours. Cathode inks were then applied in the same manner as

previously carried out and sintered at 1200 °C for 12 hours. Silver wire and silver ink

was attached to the anode and cathode layers to act as current collectors. The tubes were

again pre-reduced at 700 °C in H2/He for 1 hour.

Once again, the OCV performance was reasonable, however the performance while

recording I-V curves rapidly dropped off once in operating conditions. The electrolyte

layer applied to the hand painted samples should be considerably thicker than those

deposited using either dip coating or spray coating. This should lead to lower

performance than those other techniques due to the increased resistance to oxide

conductivity, however this increased thickness should ensure that the cell is not short

circuiting and that the anode and cathode layers remain better separated. Unfortunately,

117

the repetition of previous results remained, with cells failing to provide a constant OCV

and rapidly dropping to under 100 mV.

Heating rates were the investigated, and lowered to reduce thermomechanical stress,

from 20 °C min-1

down to 10 °C min-1

then finally 5 °C min-1

with no noticeable

difference in cell longevity. Experiments were also performed which lowered the

operating temperature to 700 °C, again with no noticeable difference on cell longevity or

performance. Fuel flow rates were also varied between 10 - 50 mL min-1

and cell tests

were carried out using both dry and wet hydrogen. Unfortunately none of these variables

managed to overcome the microtubes failing rapidly.

The aim of the extrusion work within this project was to develop a technique which

provided reproducible results to allow both the University of Birmingham to become

more self-sufficient in microtube production, as well as to also allow the direct

comparison of novel anode materials against established SOFC materials. It can then be

seen that most variables for the tube production and preparation were altered without

success. The overall shape of the extruded tubes proved difficult to control, leading to an

overall production technique which could not be thought of as repeatable or reliable.

Furthermore, the inability of the tubes to allow significant load to be drawn from them

showed major issues with the cell production. This combination of factors was

instrumental in the decision to discontinue research into microtube production and

instead focus on the synthesis and characterisation of novel materials for SOFCs. While

material synthesis made up a significant proportion of the original project goals, this shift

led to much less focus on the comparison between novel and established SOFC materials

118

using in-situ analysis and instead moved further into fundamental material

characterisation.

4.8. Further Work and Conclusions

Varying the set up of the equipment for m-SOFC testing allows small button cells to be

analysed. Fabricating cells in this geometry may allow crude comparisons of materials;

however, difficulties in repeatable layer thicknesses, cell sealing and material porosity

are problems to be overcome and all will affect cell outputs. Testing the resistance of

cells produced through the extrusion technique listed here may help to further elucidate

the cause of the poor IV performance of cells produced, which may be due to a very high

cell resistance. In addition, 3-point stress tests would assist in quantifying the mechanical

strength of the cells produced, both before and after reduction, which may help to

determine the root of the problem of this extrusion work. Alternatively, current

collection may be an issue as this is something that is a major problem in SOFC

development.

119

Chapter 5. Synthesis and characterisation of Si doped

SrFeO3-δ for both cathode and anode applications

5.1. Introduction

Perovskite materials containing transition metals have been widely investigated for

potential use with SOFCs, although the majority of the research has focused on cathode

applications. As a family of materials they show a wide range of favourable properties

from catalytic activity through to high electronic and ionic conductivity. These properties

also make them potential anode materials, which may allow the use of anodes which do

not contain nickel. This would have the potential to allow SOFCs to run directly on

hydrocarbon fuels without the need for as much fuel pre-cleaning, in particular the

removal of H2S. Several perovskite-type materials have been investigated for use as

SOFC anodes, such as Sr2MgMoO6, La0.75Sr0.25Cr0.5Mn0.5O3-δ, Sr1-xTi1-2xNb2xO3 and

others [35, 50, 60, 146–148].

Perovskite type SrFeO3-δ, has been shown to have high mixed ionic and electronic

conductivity, and has been proposed as a potential material for use in oxygen sensors,

SOFCs and other electrochemical applications [149–151]. It has been demonstrated to

have the highest conductivities when the unit cell is cubic, a phase which is only stable at

elevated temperatures [152–154]. SrFeO3-δ contains mixed valence iron cations (average

oxidation state between 3 and 4) has a range of oxygen nonstoichiometry, undergoing a

structural change leading to long-range oxygen ion vacancy ordering as iron reduces

120

towards Fe3+

[155–157]. This shift towards a brownmillerite-type structure is detrimental

for both the oxide ion and electronic conductivity [158, 159]. This work, therefore, had

the aim of stabilising the cubic form of strontium ferrate in both oxidising and reducing

atmospheres, so as to increase the overall conductivity performance.

The majority of the doping strategies for perovskite materials have focused on replacing

cations with those of similar size e.g. Mo for Cr, Co for Fe [24, 134, 160–162]. Previous

work examining doping various oxyanion species (such as carbonate, sulphate,

phosphate, borate) into perovskite systems has shown that incorporation into the

perovskite crystal structure is possible [91, 152–154, 163–167]. Another potential

oxyanion dopant is SiO44-

. In this respect, Si is often thought of as detrimental to oxygen

conductivity and cell performance within SOFC systems, as in fluorite structures such as

YSZ it collects at grain boundaries and inhibits oxide ion transport [108, 168, 169].

Relatively recently, work has been carried out to determine the effect on conductivity of

the inclusion of SiO44-

oxyanion or other tetrahedral oxyanions species within perovskite

materials to investigate the effect on performance [66, 167, 170]. In this respect, previous

work on other perovskite electrode systems has shown that doping with Si caused the

packing arrangement of SrMO3 (M=Co, Mn) to change from face-sharing octahedra to

corner-sharing octahedra, and a decrease in oxygen vacancy ordering [170]. Undoped

SrFeO3-δ already contains corner-sharing octahedra so this packing arrangement change

is not expected in this case, however the aim of this work was to inhibit the ordering of

oxygen vacancies on reduction and so increase the conductivity over the undoped

samples.

121

To this end, Si has been successfully incorporated into the perovskite structure to form

SrFe1-xSixO3-δ [171]. The work presented within this thesis investigates the effect on the

structure and conductivity of the new materials, along with electrode area specific

resistance (ASR) and thermal expansion coefficient measurements. The conductivity

behaviour in both air and reducing atmospheres was examined to determine the potential

for use as either anode or cathode SOFC materials.

5.2. Experimental

Stoichiometric quantities of high purity powders of SrCO3, Fe2O3 and SiO2 were ground

intimately before being heated to 1100 °C for 12 hours. Samples were then ball milled

(Fritsch Pulverisette 7 Planetary Mill for 1 hour at 350 rpm with 2 milling beads) and

heated to 1200 °C for a further 12 hours. Powder X-ray diffraction (Bruker D8

autosampler diffractometer using Cu Kα1 radiation) was carried out to determine phase

purity and calculate cell parameters. Unit cell parameter refinement was carried out

using GSAS suite of programs [123, 124].

For conductivity measurements, powders were again ball milled (Fritsch Pulverisette 7

Planetary Mill for 1 hour at 350 rpm with 2 milling beads) before being pressed as

pellets of 13mm diameter and sintered at 1200 °C for 12 hours. Platinum electrodes were

then attached using platinum paste (Metalor) and fired at 800 °C for 1 hour then held at

350 °C for 12 hours to ensure both bonding the Pt and full oxygenation of the samples. 4

probe D.C. conductivity measurements were then taken in both air and 5%H2/N2 at

temperatures up to 800 °C.

122

For TGA experiments, a small quantity of the sample was accurately weighed. This was

then subjected to a heating profile of 10 °C min-1

to 1200 °C in N2 followed by a holding

stage at 1200 °C for 30 minutes in order to reduce the Fe oxidation state to 3+. The

average Fe oxidation state for the sample could then be determined from the mass loss

observed.

For ASR measurements, ink suspensions were prepared containing 1:1 wt% of

electrolyte and electrode materials with 20%wt Decoflux (Zschimmer and Schwarz) to

act as a dispersant and binder. These inks were then ball milled (Fritsch Pulverisette 7

Planetary Mill for 1 hour at 350 rpm with 2 milling beads) and coated onto dense pellets

of Ce0.9Gd0.1O1.95 (CGO10, Aldrich, presintered at 1500 °C for 12 hours) to form a

symmetrical cell and then fired at 1000 °C for 1 hour. Pt paste was then applied to both

sides to act as a current collector and heated at 800 °C for 1 hour. ASR values were

obtained in both cathodic (air) and anodic (5%H2/N2) conditions at temperatures up to

800 °C (air) / 600 °C (5%H2/N2). Impedance data were collected between 5 Hz - 13 MHz

with an AC signal amplitude of 100 mV (HP 4912A Frequency Analyser). The spectra

were plotted using ZView software [127].

Samples prepared for dilatometry measurements were ball milled as powders (Fritsch

Pulverisette 7 Planetary Mill for 1 hour at 350 rpm with 2 milling beads) before being

pressed into thick pellets of 10 mm diameter and approximately 10 mm length. These

were sintered at 1200 °C for 12 hours and then polished to form smooth, flat ends.

Dilatometry data were obtained using a Netzsch 409C dilatometer, running a profile

from 25 °C to 1000 °C at 15 °C min-1

in an atmosphere of air.

123

For fuel cell testing, electrode inks were prepared with the same formula as that used for

ASR testing; 1:1 wt% of electrolyte and electrode materials with 20 wt% Decoflux

added. Pellets of ~200 μm thickness of CGO10 and Y0.15Zr0.85O2 (YSZ8, UCM

Advanced Ceramics) were prepared using a die of 13 mm diameter before being sintered

at 1500 °C for 12 hours. The 10% Si doped SrFeO3-δ ink was painted onto both sides of

these pellets and fired at 1000 °C for 1 hour, before being painted with porous silver ink

(SPI) and having silver wire attached to act as current collector. The cells were then

attached onto an alumina firing tube with internal diameter of 10 mm, external 15 mm,

using ceramabond (Aremco) and left to set for 6 hours. The firing tube was placed within

a modified tube furnace, with a length of Swagelock®

tubing running down the centre.

This tubing was used to supply the setup with hydrogen, with the inner face of the pellet

acting as the anode side of the cell and the outer face acting as the cathode side. Open

circuit voltages were obtained using a Solatron 1400A/1470E instrument, with the data

processed using the Multistat suite of software [172].

5.3. Results and Discussion

5.3.1. Synthesis and Structure Determination

Figure 26 and Figure 27 show the XRD patterns for the series SrFe1-xSixO3-δ. X-ray

diffraction data showed that incorporation of Si for Fe was possible up to 15 % mol;

above this level small Sr2SiO4 peaks appeared showing that the lattice solubility limit

124

had been exceeded. It can also be seen that the unit cell changes from tetragonal to cubic

upon doping with Si which corresponds to the disordering of oxygen vacancies.

Si mol % 0 5 10 15

a (Å) 10.9235 3.8636 3.8723 3.8755

c (Å) 7.6965 = a = a = a

Volume (Å3)

per formula unit

57.40 57.67 58.06 58.34

Table 9. Unit cell parameters and volumes (per formula unit) for the series SrFe1-xSixO3-δ

Figure 26. XRD pattern for SrFeO3-δ in air (bottom) and annealed in 5%H2/N2 at 600 °C

for 24 hours (top)

125

Figure 27. XRD pattern for SrFe0.9Si0.1O3-δ in air (bottom) and annealed in 5%H2/N2 at


Table 9 displays the unit cell volumes per formula unit for the series SrFe1-xSixO3-δ. A

trend of increasing cell volume with increasing Si content was noticed. This has also

been previously reported for Si doping in Mn perovskites [170, 173] but is unusual as

Si4+

is smaller than either the Fe3+

or Fe4+

it replaces, with tetrahedral Si having an ionic

radius of 0.26 Å against octahedral Fe3+

at 0.645 Å and Fe4+

at 0.585 Å [173, 174]. This

increase in cell parameter is attributed to the introduction of Si leading to a reduction of

Fe4+

to Fe3+

. This can be explained by the strong tetrahedral coordination preference of

Si which ensures the introduction of oxygen vacancies and hence partial reduction of

iron according to the Krőger-Vink notation equation below:

126

SiO2 + 3FeFeX +OO

X SiFe

X + 2FeFe’ +VO

•• + ½ O2 + “FeO2”

Figure 28. Kröger-Vink notation showing silica incorporation causes oxygen vacancies

and Fe reduction

It should also be noted that it has been similarly reported in the literature that

incorporation of a cation with a smaller or similar ionic radius than iron, such as Al3+

,

Ga3+

and Mo6+

, also leads to an increase in the unit cell size, and this was also attributed

to the lowering of the proportion of Fe4+

[175, 176].

Si mol % 0 5 10 15

Fe Oxidation State 3.80 3.60 3.48 3.42

Oxygen Deficiency (δ) 0.10 0.20 0.23 0.25

Table 10. Oxidation state and oxygen deficiencies for the series SrFe1-xSixO3-δ

Confirmation of the reduction in the average iron oxidation state was shown by TGA

studies. This was carried out by accurately weighing a sample before heating at 10°C

min-1

to 1200°C in N2, and holding for 30 minutes, in order to reduce the iron present to

Fe3+

. The weight loss experienced by the sample is attributed to the loss of oxygen and

can be used to calculate the original oxidation state of the iron. The values obtained are

shown in Table 10. It was found that as the Si content is increased the number of oxygen

vacancies increased and the average oxidation state of iron decreased.

127

In order to examine the stability in strongly reducing conditions, XRD patterns of

samples were recorded after reduction at 600 °C in 5% H2/ 95% N2. This showed that the

undoped SrFeO3-δ changed structure from tetragonal to the brownmillerite-type

orthorhombic Sr2Fe2O5, with contains ordered oxide ion vacancies (see Figure 26). This

was in contrast to the doped samples which remained as cubic systems, indicating that

disordered vacancies were still present. Thus Si doping has been demonstrated to

increase long range oxide ion vacancy disorder under both oxidising and reducing

conditions.

5.3.2. Electrical Conductivity Measurements

Electrical conductivity measurements were taken for several SrFe1-xSixO3-δ samples ( 0 ≤

x ≤ 0.15) using 4-probe DC conductivity analysis. Measurements were carried out in

both air and 5%H2/N2 to examine the potential for use as cathode and anode SOFC

materials respectively.

5.3.2.1. Air Atmosphere

Conductivity measurements were carried out on dense sintered pellets of x = 0, 0.05, 0.1,

0.125 and 0.15 and the results are shown in Figure 29 and Table 11. It was observed that

conductivity was highest for the x = 0.1 sample at 24.1 S cm-1

at 800 °C and 35.3 S cm-1

at 700 °C, with both the higher and lower Si doped samples recording lower values. The

initial improved conductivity on Si doping may either be attributed to the change from a

128

tetragonal to a cubic system, leading to an improvement in conduction pathways, or be

due to changes in the Fe oxidation state.

Figure 29. Plot of conductivity against temperature for the series SrFe1-xSixO3-δ in air

Si mol % 0 5 10 15

800 °C (S cm-1

) 17.2 14.6 24.1 13.4

700 °C (S cm-1

) 26.3 21.2 35.3 17.8

600 °C (S cm-1

) 37.2 28.8 49.0 21.4

500 °C (S cm-1

) 47.9 36.3 58.8 23.4

400 °C (S cm-1

) 45.7 45.4 67.4 22.4

Table 11. Conductivity data for SrFe1-xSixO3-δ in air

0

10

20

30

40

50

60

70

80

650 750 850 950 1050

σ /

Scm

-1

T (K)

x = 0

x = 0.05

x = 0.1

x = 0.15

129

As the Si doping is increased further, the lowering of the conductivity is postulated as

being due to the increasing concentration of the silicate oxyanion interrupting the Fe-O

conduction network. All samples showed an increase in conductivity to 400 °C, a

relatively stable conductivity in the temperature range 400 - 600 °C, and a decrease at

higher temperatures. The loss of oxygen above 400 °C is most likely accountable for this

drop in conductivity above these temperatures as the average oxidation state of Fe is

reduced, although it is partially counterbalanced through the semiconductor behaviour

leading to increasing hole mobilities with increasing temperatures. Overall, these

conductivities compare favourably with literature examples, for example LSCM has been

shown to have total conductivity at 900 °C in air of 38.6 S cm-1

[49].

5.3.2.2. 5%H2/N2 Atmosphere

Table 12 and Figure 30 display the conductivity data for the series under 5% H2/N2. In

these conditions, the reduction of the average oxidation state of iron towards Fe3+

above

400 °C causes a large drop in the conductivities of the samples analysed. Once again, the

x = 0.1 sample shows the highest level of conductivity with the lowest results posted for

x = 0 and 0.05 samples. This in part can be attributed to the x = 0 and 0.05 samples not

being fully stabilised cubic systems and so forming the brownmillerite-type phase (SrFe1-

xSixO2.5) on reduction. On increasing the temperature further, an increase in conductivity

was observed for all samples, which may result from mixed Fe3+

/Fe2+

valence, although

the conductivities were still lower than the values in air. This behaviour can be seen in

literature examples of MIEC systems such as Sr2Fe1.5Mo0.5O6-δ [177].

130

Figure 30. Plot of conductivity against temperature for the series SrFe1-xSixO3-δ in

5%H2/N2

Si mol % 0 5 10 15

800 °C (S cm-1

) 1.6 0.7 11.3 3.8

700 °C (S cm-1

) 0.9 0.5 8.3 3.0

600 °C (S cm-1

) 0.6 0.3 5.4 2.7

500 °C (S cm-1

) 0.1 0.2 3.3 3.0

400 °C (S cm-1

) 0.1 0.1 1.9 3.8

Table 12. Conductivity data for SrFe1-xSixO3-δ in 5%H2/N2

In this respect, it is interesting to note that doped samples that were held at 800 °C for 30

minutes, before having conductivity measurements taken on decreasing temperature,

0

2

4

6

8

10

12

650 750 850 950 1050

σ /

Scm

-1

T (K)

x = 0

x = 0.05

x = 0.1

x = 0.15

131

showed significantly higher electronic conductivities than the samples measured on

increasing temperatures. This may be due either to the partial reduction of Fe3+

to Fe2+

or

to the precipitation of iron metal out of the structure which could be providing additional

conducting species (as seen in literature examples involving Ni and Ru), although there

was no clear evidence for this by XRD [178]. Again the x = 0.1 sample displayed the

highest conductivity, reaching 11.34 S cm-1

at 800 °C and displaying a large drop on

decreasing the temperature. The undoped sample showed little variation in conductivity

and did not display the same large drop with decreasing temperature.

Overall the lower conductivities under 5% H2/N2 than air may hinder the use of Si-doped

strontium ferrites as a stand-alone anode material; however they may function well as

cermets with nickel or other metals. The increase in disordering of the oxygen vacancies

upon the introduction of silicon should result in enhanced oxide ion conductivity over the

undoped sample and hence further increase their suitability as anode cermet oxides.

5.3.3. Area Specific Resistance

Area specific resistance studies were carried out on the series SrFe1-xSixO3-δ in both air

and 5% H2/N2 atmospheres. As described above, a composite of the perovskite and

CGO10 was deposited on dense CGO10 pellets, and fired at 1000 °C for 1 hour before

having electrodes attached.

132


Si mol % 0 5 10 15

700 °C ASR (Ω cm2) 1.65 1.19 0.90 1.88

800 °C ASR (Ω cm2) 0.25 0.14 0.08 0.28

Table 13. ASR values for the series SrFe1-xSixO3-δ in air

Figure 31. Nyquist plot for SeFe0.9Si0.1O3- δ at 800 °C in air, including fitting

Values for ASR in air can be seen in Table 13 and a Nyquist plot for the 10% Si sample

at 800°C is shown in Figure 31. A considerable improvement was obtained for the 5 and

Z' (Real Impedance) / Ω

133

10% mol doped samples over the undoped SrFeO3-δ, with values of 0.14 and 0.08 Ω cm2

respectively against 0.25 Ω cm2 at 800 °C. Once again, the 15% mol doped sample

displayed a drop in performance from that of both the 10% mol doped and undoped

sample, with a value of 0.28 Ω cm2 at 800 °C. This may be due to the disruption of the

electron conductivity pathway or the possible presence of amorphous Si-containing

impurities at this higher Si level, causing increased resistances. The loss of oxygen at

higher temperatures may explain the significant improvement in ASR at elevated

temperatures; loss of oxygen leads to an increase in oxide ion mobility through the

increase in the concentration of oxide ion vacancies.

It should be noted that a perfect semicircle is not formed by the material, as represented

on the Nyquist plot. This is due to the different resistances of the bulk material and the

grain boundary, with the two semicircles representing each being difficult to distinguish.

However, as both parts in combination represent the overall resistance, the intercept

points still hold true.

5.3.3.2. 5%H2/N2 Atmosphere

Si mol % 0 5 10 15

600 °C ASR (Ω cm2) 1.88 1.51 0.17 0.24

Table 14. ASR values for the series SrFe1-xSixO3-δ in 5%H2/N2

134

Figure 32. Nyquist plot for SeFe0.9Si0.1O3- δ at 600 °C in 5%H2/N2, including fitting

Values for ASR against temperature in 5%H2/N2 can be seen in Table 14. As before, an

improvement was noticed for the 5 and 10% mol doped samples over the undoped

SrFeO3-δ, with values of 1.51 and 0.17 Ω cm2 against 1.88 Ω cm

2 at 600 °C. While the

15% mol doped sample displayed a drop in performance compared to the 10% mol

doped sample, but it was significantly improved compared to the 5% mol doped and

undoped samples, with a value of 0.24 Ω cm2 at 600 °C. The reduced performance of the

5% Si mol doped sample against the 15% mol doped sample may be due to the lower Si

content not fully stabilising the cubic phase and on reduction, allowing the ordered

Sr2Fe2O5 phase to partially form. The ASR performance was seen to improve at higher

temperatures in air, for example the 10% mol doped sample had values of 0.90 vs 0.08 Ω


135

cm2 for 700 vs 800 °C. If this was to hold true for reducing atmospheres then the

performance of the 10 and 15% mol doped materials may far surpass the air performance

based on the figures obtained at 600 °C. However, since the Ce within the CGO10

electrolyte reduces above 600 °C, such higher temperature measurements in 5% H2/N2

could not be performed and so further studies are warranted analysing the performance

of these systems utilising a YSZ electrolyte.

Again, as can be seen in Figure 32, two separate curves contribute to the overall

resistance of the sample, with the grain boundary and the bulk resistances being difficult

to distinguish from one another.

5.3.4. Dilatometry

Dilatometry studies were carried out on the series SrFe1-xSixO3-δ with x = 0, 0.05 and

0.10 in air. The results are shown in Table 15 and Figure 33, Figure 34 and Figure 35.

Si mol % 0 5 10

TEC in temperature band 20 - 480 °C (x 10-6

/K) 19.3 18.6 14.8

TEC in temperature band 480 - 1000 °C (x 10-6

/K) 37.4 34.5 32.3

Table 15. TEC data for the series SrFe1-xSixO3-δ

136

Figure 33. Dilatometry data for SrFeO3-δ from 20 - 1000 °C at 15 °C min-1

in air. Solid

line is TEC profile, dashed line is temperature profile


in air.


137


in air.


As expected, a reduction in the thermal expansion coefficient (TEC) is seen in the doped

sample versus the undoped; this has previously been reported to occur for other doped

SrFeO3-δ series, such as SrFe1-xMxO3-δ (M = Ga, Al, Ti) and other related materials [175,

179–181]. The TECs recorded here are in line with literature values for these other doped

strontium ferrite series. Both the doped and undoped dilatometric curves display a non-

linear behaviour above approximately 480 °C, which may be attributed to the loss of

oxygen and subsequent reduction of Fe4+

to the larger Fe3+

causing an increase in the

TEC. The reduction in TEC on Si doping can then be related to the lower average Fe

oxidation state, as on heating the level of oxygen loss and hence reduction is lower. The

observed reduction in TEC for the 10% Si doped sample over the undoped sample from

19.3 to 14.8 x 10-6

/K below 480 °C, and from 37.4 to 32.3 x 10-6

/K above 480 °C,

138

would assist in making SrFe0.9Si0.1O3-δ a more suitable SOFC material as it has a closer

match to the TEC of existing SOFC materials. However the higher temperature TECs are

still probably too high, limiting SrFe0.9Si0.1O3-δ to lower temperature use. Alternatively,

by utilising a composite with the electrolyte material the difference in TEC values could

be bridged for use at higher temperatures. Overall, the data suggests that intermediate

temperature operation in the region of 500 - 600 °C would be a suitable use for the 10%

Si material alone as it would take maximum advantage of the lower TEC while still

being at temperatures high enough to allow internal reforming of hydrocarbon fuels.

Future work focussing on detailing the precise temperature of the loss of oxygen would

be of interest. This could be achieved through having a slower rate of heating between

450 - 500 °C, for example, which would give much more accurate information as to

when the loss occurs, allowing more details about the material to be documented.

5.3.5. Fuel Cell Testing

A symmetrical pellet cell [79] of SrFe0.9Si0.1O3-δ was prepared and run as per Chapter 3,

on pellets of electrolyte material formed using a 13mm die set. The internal diameter of

the firing tube that the pellets were cemented onto was 12mm, giving an active area of

approximately 1.1 cm2. The test setup can be seen in Figure 11. The initial results

appeared promising for novel materials, with little gas leaking or short circuiting of the

cell evident and OCVs in the region of 0.9 V. Unfortunately on all cell tests the OCV did

not translate to a reasonable cell performance, with the voltage dropping rapidly when

attempting to draw a current from the cell, and also the OCVs dropping off with 30

139

minutes, down to less than 100 mV. This behaviour was very similar to the cells

fabricated in Chapter 4. Examples of cell voltages obtained from attempts to draw power

from the cell can be seen in Figure 36.

Figure 36. Voltages and currents for symmetrical pellet cells of SrFe0.9Si0.1O3-δ on YSZ at

800°C moving from OCV to 0.85 V voltage demand

The testing profile for the results seen in Figure 36 involved running the cell at OCV

before attempting to draw a voltage of 0.85 V. Data were logged at 5 Hz. The settings

used were chosen due to the lower OCVs displayed by the cells. This was speculated to

be due to the cells being electrolyte-supported, causing higher ohmic resistances than

cells with thinner electrolyte layers. Again, on attempting to draw power from the cell,

the voltage falls below the set-point without significant current being produced.

However, as the cells are electrolyte supported, questions over cells short-circuiting

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.005 0.01 0.015 0.02 0.025 0.03

Vo

lta

ge

(V)


Cell 1

Cell 2

Cell 3

140

should be alleviated, and achieving OCVs over 0.9 V also indicates little gas crossover.

Further work is needed to optimise the cell production to obtained improved, repeatable

performance, as per Chapter 4.

5.4. Conclusions

In this work it has been shown that Si can be successfully incorporated into the

perovskite structure. The solubility limit for SrFe1-xSixO3-δ was found to be x = 0.15;

above this level of Si, impurities were detected. The 5% mol Si doped sample did not

fully stabilise the cubic phase in a reducing atmosphere. The presence of tetrahedral

silicate oxyanions has therefore been shown to assist in stabilising oxygen vacancy

disorder in SrFeO3-δ at higher temperatures, which has a positive influence on the

electrochemical performance of the series leading to a lowering of area specific

resistance. In addition, the thermal expansion coefficient for the doped samples under

500 °C was closer to that of many standard SOFC materials, possibly facilitating easier

incorporation, although at temperatures significantly above this it was still too high with

regards to conventional fuel cell electrolytes.

Work within this Chapter was carried out in conjunction with Dr Jose Porras and has

been published in the Journal of Materials Chemistry A, Volume 3, Issue 38, Pages

11834 - 11841, 2013.

141

Chapter 6. Synthesis and characterisation of V doped

Sr0.8Ti0.6Nb0.4O3-δ for SOFC Applications

6.1. Introduction

Replacing elements within known structures with small quantities of other elements is a

standard approach for altering or enhancing material properties. Within the family of

perovskite materials, A-site deficient compounds are one subcategory of materials, many

of which have gained attention in part due to promising thermal stability, electrical

characteristics and chemical compatibility with other materials potentially used within

SOFC systems. Examples of A-site deficient perovskite materials include

(La0.75Sr0.25)0.95Cr0.5Mn0.5O3-δ (LSCM) (Sr1-xLax)1-yTi1-zO3 and La0.4Sr0.4GaxTi1-xO3-x/2

amongst others [39, 42, 44, 182, 183]. Synthesising A-site deficient perovskite materials

can be achieved very simply, as in the case of LSCM, by using non-stoichiometric

quantities of reactants. However, it is also possible to form deficiencies though doping

with one or more cations of a different charge to the cation they are replacing. Such work

often centres on charge balancing the crystal structure by doping on both the A- and B-

sites. Doping with species of different charge in order to form A-site deficient materials

has been well documented in the past with series such as the perovskites Sr1-x/2Ti1-

xNbxO3-δ, Sr0.86Y0.08TiO3-δ and others [35, 38, 135, 146, 184, 185]. The presence of A-site

vacancies can also in turn promote oxygen vacancies which improve oxide transportation

and so can assist in the performance of MIEC materials, such as the double perovskite

material series La2-xNiTiO6-δ [186].

142

In addition, the related tungsten bronze phase Sr0.6Ti0.2Nb0.8O3-δ and its variants has been

investigated after it was identified as an impurity in the perovskite system Sr1-x/2Ti1-

xNbxO3-δ if x > 0.4 [45]. These and other related tungsten bronze materials displayed

promising conductivities but did show minor chemical compatibility issues with

commonly-utilised electrolyte material YSZ at elevated temperatures [47].

However, both Sr1-x/2Ti1-xNbxO3-δ and Sr0.6Ti0.2Nb0.8O3-δ display Warburg type behaviour

when studied using AC impedance, implying slow oxygen transport due to a diffusion

limited process [47]. Investigation into this and other series have confirmed that this

diffusion limited process is also at fault for the slow reoxidisation of reduced samples by

studying porous samples which do not display the same behaviour [187]. While oxygen

diffusion-limited processes were shown to be possible to avoid by creating a much more

porous sample [188] it still will impede the performance within a SOFC by reducing the

available triple phase boundary as slow oxygen ion mobility impedes the overall

electrochemical reactions. The tungsten bronze-type series displayed improved

reoxidisation kinetics and hence inferred greater oxide ion conduction than the

perovskite series; however, the level was still comparatively low for SOFC electrode

materials. There is consequently a need to investigate strategies to improve the oxide ion

conductivity.

Doping with vanadium has been shown to increase the catalytic activity towards steam

reforming within perovskite anode materials while reducing carbon deposition [189] and

is more stable in its lower oxidation states than niobium which may assist in increasing

the performance of a material doped with vanadium on the niobium site in a reducing

143

atmosphere [37, 190]. In particular, lowering the oxidation state of vanadium upon

reduction would promote oxygen loss and may improve oxide ion conductivity and

redox properties at SOFC operating temperatures. For these reasons, V doping into

Sr0.8Ti0.6Nb0.4O3-δ was investigated, forming Sr0.8Ti0.6Nb0.4-xVxO3-δ, as this particular

material was shown to be the limit of niobium doping in Sr1-x/2Ti1-xNbxO3-δ while

retaining the perovskite structure, with the tungsten bronze phase present as an impurity

at higher levels. This large incorporation of Nb within the perovskite unit cell allows for

the greatest range in V doping and so the greatest opportunity to study the effects upon

the material characteristics. During the course of these vanadium doping studies, it was

shown that the unit cell was also able to accommodate a greater proportion of vanadium

than niobium, with phase pure Sr0.75Ti0.5V0.5O3-δ being successfully synthesised and

analysed. In addition, limited synthesis and analysis of the tungsten bronze series

Sr0.6Ti0.2Nb0.8-xVxO3 was carried out in order to determine the effect of doping vanadium

into this system.

6.2. Experimental

Stoichiometric quantities of high purity powders of SrCO3, Nb2O3 and V2O5 were ground

intimately before heating to 1100 °C for 12 hours. Samples were then ball milled (Fritsch

Pulverisette 7 Planetary Mill for 1 hour at 350 rpm) and heated to 1300 - 1400 °C for a

further 12 hours. Powder X-ray diffraction (Bruker D8 autosampler diffractometer using

Cu Kα1 radiation) was carried out to ensure phase purity and calculate cell parameters.

Unit cell refinement was carried out using GSAS and EXPGUI [123, 124].

144

For 4-probe DC conductivity measurements, powders were ball milled (Fritsch


pressed as pellets of 13mm diameter and sintering at 1350 °C for 12 hours. Platinum

electrodes were then attached using platinum paste (Metalor) and fired at 800 °C for 1

hour to bond the Pt. 4 probe DC conductivity measurements were then taken in H2/N2

ascending from 300 °C to 800 °C, before reducing at 930 °C for 1 hour and then taking

measurements down from 800 °C to 300 °C.

For ASR measurements, ink suspensions were prepared containing 1:1 wt% of

electrolyte and electrode materials with 20 %wt Decoflux (Zschimmer and Schwarz) to

act as a dispersant and binder. Samples were then ball milled (Fritsch Pulverisette 7

Planetary Mill for 1 hour at 350 rpm with 2 milling beads) and coated onto dense

Zr0.85Y0.15O1.925 (YSZ8, UCM Advanced Ceramics) pellets (sintered at 1500 °C for 12

hours) to form a mirrored cell and then fired at 1000 °C for 1 hour. Pt paste was then

applied to both sides to act as a current collector and bonded at 800 °C for 1 hour. ASR

values were obtained in both cathodic (air) and anodic (5%H2/N2) conditions at

temperatures up to and 800 °C. Impedance data were collected between 5 Hz - 13 MHz

at open circuit voltage (OCV) with an A.C. signal amplitude of 100 mV (HP 4912A

Frequency Analyser). The spectra was plotted using ZView software to give an estimate

of the resistance and capacitance of the different cell component materials [127].

Samples prepared for dilatometry measurements were ball milled as powders (Fritsch


pressed into thick pellets of 10 mm diameter and approximately 10 mm length. These

145

were sintered at 1350 °C for 12 hours and then polished to form smooth, flat ends.

Dilatometry data were obtained using a Netzsch 409C dilatometer, running a profile

from 25 °C to 1000 °C at 15 °C min-1

in an air atmosphere.



The unit cell for Sr0.8Ti0.6Nb0.4O3-δ has previously been shown to be cubic and all data

obtained here confirms this [146]. X-ray diffraction data showed that it was possible to

completely replace Nb5+

with V5+

within the perovskite structure for Sr0.8Ti0.6Nb0.4-

xVxO3-δ, and phase pure samples were prepared for values of x between 0 and 0.4 at 0.05

intervals. In addition, the V level and A-site deficiency was further increased by

increasing the vanadium doping to give Sr0.75Ti0.5V0.5O3-δ without impurities being

detected. Figure 38, Figure 39, Figure 40 and Figure 41 show a selection of XRD

patterns for this series. These are clean, with well-defined peaks, showing no impurities

present.

146

x a (Å) V (Å3)

0 3.9298(1) 60.690(1)

0.05 3.9237(1) 60.406(1)

0.1 3.9192(1) 60.200(1)

0.15 3.9122(1) 59.876(1)

0.2 3.9110(1) 59.820(1)

0.25 3.9090(1) 59.732(1)

0.35 3.9019(1) 59.407(1)

0.4 3.9008(1) 59.357(1)

Table 16. Unit cell size and cell volume for Sr0.8Ti0.6Nb0.4-xVxO3-δ

Figure 37. Graph showing unit cell volume (Å3) for Sr0.8Ti0.6Nb0.4-xVxO3-δ against V

content

59.2

59.4

59.6

59.8

60

60.2

60.4

60.6

60.8

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45

un

it c

ell

volu

me

(Å3)

value of x

147





148

Figure 40. XRD pattern for Sr0.8Ti0.6V0.4O3-δ in air (bottom) and annealed in 5%H2/N2 at


Figure 41. XRD pattern for Sr0.75Ti0.5V0.5O3-δ in air showing single phase purity

149

Cell parameters for the unreduced series are shown in Table 16. As niobium is replaced

by vanadium the unit cell size decreases. This is fully expected as the V5+

cation is

smaller than Nb5+

(0.54 versus 0.64Å) which would predict a reduction in unit cell size

on V incorporation, and even if vanadium is reduced down to V4+

(0.58Å) the unit cell is

still expected to contract [174]. The one anomalous point at x = 0.3 may be caused by

very small levels of impurities being present in this particular phase.

The sample Sr0.75Ti0.5V0.5O3-δ showed an increased unit cell size over the Sr0.8Ti0.6V0.4O3-

δ sample, with values of a = 3.9071Å and V = 59.644Å3. This increase was not expected

as the V5+

cation is smaller than the Ti4+

it is replacing (0.54 versus 0.605Å). In addition,

the increase in A-site vacancies necessitated by the need for charge balancing the

replacement of Ti4+

with V5+

might be expected to bring about a further reduction in unit

cell size. Further studies are required to investigate the origin of this increase, but it may

relate to the influence of some partial A-site vacancy ordering as a result of the high

levels of vacancies.

A selection of phase pure materials from the series was reduced at 930 °C for 24 hours in

an atmosphere of 5%H2/N2. The unit cell size and volumes calculated for several

samples can be seen in Table 17 and Figure 42, and Figure 38, Figure 39, Figure 40 and

Figure 41 display XRD patterns for several reduced species against their unreduced

counterparts.

150

x a (Å) V (Å3)

0.15 3.9030(1) 59.456(1)

0.2 3.9050(1) 59.547(1)

0.25 3.9071(1) 59.644(1)

0.4 3.9018(1) 59.401(1)

Table 17. Unit cell size and cell volume for Sr0.8Ti0.6Nb0.4-xVxO3-δ reduced at 930 °C in

5%H2/N2

Figure 42. Graph showing unit cell volume (Å3) for Sr0.8Ti0.6Nb0.4-xVxO3-δ (unreduced and

reduced at 930 °C) against V content

For the samples x = 0.15, 0.2 and 0.25, the unit cell size increases with increasing

vanadium content which can be explained by the reduction of some of the V5+

counteracting the size difference between Nb5+

and V5+

(0.64Å vs 0.54Å), with

59.2

59.3

59.4

59.5

59.6

59.7

59.8

59.9

60

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45

un

it c

ell

volu

me

(Å3)

value of x

reduced at 930°C

unreduced

151

vanadium reducing more readily than niobium. However, when vanadium fully replaces

niobium it can be seen that the unit cell size decreases again. It should be noted that the

only sample in which the unit cell volume of the reduced sample is larger than that of the

unreduced sample is x = 0.4. This runs contrary to expectations as it would be anticipated

that the reduction of Nb5+

/V5+

to Nb4+

/V4+

would lead to an expansion in unit cell due to

the increase in cationic radius, and directly contradicts previous findings on the series

Sr1-x/2Ti1-xNbxO3-δ showing an increase in unit cell size on reduction [35, 146, 187]. It

may be that due to the larger reduction, the larger level of oxide ion vacancies (which

contribute to a cell size reduction) may balance the effect of the larger V4+

cation size. In

addition, it should be noted that on closer inspection, all of the spectra for the reduced

samples contain a very small secondary shoulder peak which can be seen to be forming

at a slightly higher 2θ value than the large peak at ~32° which is not present on the

undoped sample, possibly indicating the partial breakdown of the perovskite structure or

some degree of oxide ion vacancies ordering, for example forming tetrahedral clusters

around the vanadium cation. Unfortunately this peak is too small to gather any further

detailed information.

6.3.2. 4-Probe DC Conductivity

4 probe DC conductivity measurements were carried out on the series Sr0.8Ti0.6Nb0.4-

xVxO3-δ (≤ x ≤ 0.4), with the focus on conductivities in a reducing atmosphere. To

achieve this, samples were reduced at 930 °C for 30 minutes before analysis. In addition,

a sample of Sr0.75Ti0.5V0.5O3-δ was also analysed. Figure 43 and Table 18 display

conductivities of samples for a range of temperatures. As anticipated, the unreduced

152

samples displayed very low conductivities in air. This trend was also shown at lower

temperatures temperature in 5%H2/N2. Sr0.8Ti0.6Nb0.2V0.2O3-δ was shown to have the

highest consistent conductivities of the group with values of 2.34 and 5.63 S cm-1

at 800

°C and 300 °C respectively.

Figure 43. Plot of conductivity against temperature for the series Sr0.8Ti0.6Nb0.4-xVxO3-δ

in 5%H2/N2

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

650 750 850 950 1050

σ /

Scm

-1

T (K)

x = 0

x = 0.05

x = 0.1

x = 0.15

x = 0.2

x = 0.25

x = 0.3

x = 0.35

x = 0.4

153

x 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

800 °C (S cm-1

) 0.46 0.34 0.15 1.10 2.34 1.16 0.79 0.04 0.04

700 °C (S cm-1

) 0.42 0.28 0.09 1.10 2.58 1.20 0.87 0.02 0.05

600 °C (S cm-1

) 0.33 0.21 0.05 1.20 3.00 1.33 1.03 0.02 0.06

300 °C (S cm-1

) 0.18 0.12 0.00 1.79 5.63 2.61 2.16 0.01 0.11

Table 18. 4-probe DC conductivity data for the series Sr0.8Ti0.6Nb0.4-xVxO3-δ in 5%H2/N2

post reduction at 930 °C

Upon increasing proportion of vanadium, the general conductivity behaviour moved

from that of a semiconductor-like material to that of a metal-like material, albeit showing

low electronic conductivity. This was indicated by the tendency for metallic materials to

display a reduction in conductivity with increasing temperature and for the reverse to

hold true for semiconductors. This behaviour may be attributed to the reduction of V5+

to

V4+

. However, above x = 0.3 the conductivity drops to a very low level, with the x = 0.35

and 0.4 samples along with the Sr0.75Ti0.5V0.5O3-δ sample displaying values at least one

order of magnitude lower than the 0.2 sample.

6.3.3. Area Specific Resistance

Since Sr0.8Ti0.6Nb0.2V0.2O3-δ was shown to have the highest conductivity in reducing

conditions, a sample was investigated further as an anode material. A layer of

Sr0.8Ti0.6Nb0.2V0.2O3-δ was coated onto a dense pellet of 8%mol YSZ. Values for ASR

154

were recorded at 700 and 800 °C in both air and 5%H2/N2, with the latter being recorded

post reduction at 930 °C. The results are shown in Table 19 and Nyquist plots for

samples analysed at 800°C in both air and 5% H2/N2 are in Figure 44 and Figure 45.

5%H2/N2 Air

800 °C (Ω cm2) 1.20 16.16

700 °C (Ω cm2) 2.93 60.23

Table 19. ASR data for the material Sr0.8Ti0.6Nb0.2V0.2O3-δ in both air and 5%H2/N2

As expected, the performance of the material in air is far worse than that in 5%H2/N2;

however, neither of the conditions produced particularly favourable results with respect

to conventional anode materials, with the value of 1.2 Ω cm2 being about an order of

magnitude higher than a Ni/YSZ composite. As it can be seen, the drop in performance

at lower temperatures is much more pronounced in the air samples.

155

Figure 44. ASR Nyquist plot for Sr0.8Ti0.6Nb0.2V0.2O3-δ at 800 °C in air, including fitting

As in Figure 31 and Figure 32 in Chapter 5, the bulk and grain boundary resistances can

both be seen contributing to the overall resistance. However, it is interesting to note that

the grain boundary resistance appears more pronounced in the sample analysed under

reducing conditions, suggesting that either the effect of grain boundary resistance is

increased on reduction or that the bulk resistance is decreased by a larger margin under

reducing conditions.


156

Figure 45. ASR Nyquist plot for Sr0.8Ti0.6Nb0.2V0.2O3-δ at 800 °C in 5%H2/N2, including

fitting

6.3.4. Reoxidation Studies

Samples of Sr0.8Ti0.6Nb0.4-xVxO3-δ as prepared for 4-probe DC conductivity were reduced

at 930 °C for 30 minutes before being held at 800 °C. The gas supply was then isolated

and the furnace chamber vented to the atmosphere to allow air inside. The conductivity

of samples was then measured over a period of up to 25 minutes to ascertain the

reoxidisation of the samples, as all samples showed very low conductivity before


157

reduction. This study was carried out to determine the ability of the sample to reoxidise,

giving an indication of the ability of oxygen and oxide ions to diffuse through the

sample. If reoxidisation occurs slowly, as with previous studies, a Warburg-type

behaviour may again be suspected as responsible [47].

x

time

0 0.05 0.1 0.15 0.2 0.35 0.4

0s 0.426 0.313 0.149 1.096 2.269 0.514 0.046

60s 0.423 0.311 0.151 0.485 2.100 0.149 0.039

120s 0.424 0.021 0.010 0.109 0.515 0.004 0.000

300s 0.426 0.001 0.000 0.000 0.040 0.000 0.000

Table 20. Conductivity (S cm-1

) against time (s) for samples of Sr0.8Ti0.6Nb0.4-xVxO3-δ

exposed to air post reduction in 5%H2/N2

Table 20 and Figure 46 shows a selection of conductivities against time for selected

samples of Sr0.8Ti0.6Nb0.4-xVxO3-δ exposed to air post reduction. As can be seen, all

samples containing vanadium rapidly dropped towards negligible conductivity once

exposed to an oxidising environment. This clearly contrasts with the vanadium-free

sample of Sr0.8Ti0.6Nb0.4O3-δ which remained at a fairly consistent conductivity despite

the change in environment. It should be noted that the conductivity of this undoped

sample remained at 0.417 S cm-1

after 30 minutes of exposure to air at 800 °C, however

the drop recorded may be due to temperature fluctuations caused by opening the furnace

chamber to air.

158

Figure 46. Conductivity against time for samples of Sr0.8Ti0.6Nb0.4-xVxO3-δ exposed to air

at 800 °C post reduction in 5%H2/N2 at 930 °C

While these data cannot be used to directly measure oxygen transfer, it can be inferred

that oxygen transfer is improved in the doped samples versus the undoped. However, the

level of doping appears to have little influence on the rate of reoxidisation, with all

vanadium-containing samples dropping several orders of magnitude within 5 minutes.

6.3.6. Dilatometry

Dilatometry studies were carried out on both Sr0.8Ti0.6Nb0.4O3-δ and Sr0.8Ti0.6Nb0.2V0.2O3-

δ in air. The results are shown in Table 21, Figure 47 and Figure 48.

0

0.5

1

1.5

2

2.5

0 50 100 150 200 250 300 350

Con

du

ctiv

ity (

S c

m-1

)

Time (s)

x = 0

x = 0.05

x = 0.1

x = 0.15

x = 0.2

x = 0.35

x = 0.4

159

Sample TEC (x10-6

/K)

Sr0.8Ti0.6Nb0.4O3-δ 11.73

Sr0.8Ti0.6Nb0.2V0.2O3-δ 12.65

Table 21. TEC data for the materials Sr0.8Ti0.6Nb0.4O3-δ and Sr0.8Ti0.6Nb0.2V0.2O3-δ

Figure 47. Dilatometry data for Sr0.8Ti0.6Nb0.4O3-δ from 20 - 1000 °C at 15 °C min-1

in


160

Figure 48. Dilatometry data for Sr0.8Ti0.6Nb0.2V0.2O3-δ from 20 - 1000 °C at 15 °C min-1

in


As can be seen, the TEC is larger for the vanadium doped samples than for the undoped.

This is in line with expectations, as the vanadium-containing sample is much more prone

to losing oxygen upon heating, leading to a slight average increase in B-site cation

radius.

Both samples displayed a very slight increase in the rate of expansion above ~700 °C

which may be attributed to the loss of oxygen leading to the reduction (and subsequent

increase in radius) of the Nb/V on the B site. However, this increase in expansion rate is

small, suggesting limited reduction in air.

161

6.4. Conclusions and Further Work

This work has demonstrated that it is possible to replace niobium with vanadium in the

perovskite material Sr0.8Ti0.6Nb1-xVxO3-δ while retaining phase purity. The inclusion of

some vanadium improves the reoxidisation behaviour and the conductivity in reducing

conditions over the undoped sample; however, over x = 0.2 the performance decreased.

Thermal expansion coefficients are a close match for existing SOFC materials such as

YSZ (10.5 x 10-6

/ K); however, further work needs to be carried out to ascertain the

chemical compatibility with such systems. Additional work investigating V doping of

other Sr1-x/2Ti1-xNbxO3-δ systems, or doping into Sr1-3x/2LaxTiO3, may also give further

insight into some of the unexpected characteristics of Sr0.8Ti0.6Nb0.4-xVxO3-δ. In-situ

studies of these materials as part of a solid oxide fuel cell should also provide valuable

insights into material performance.

162

Chapter 7. Synthesis and characterisation of La1-

xMnxMn1-xTixO3-δ for SOFC applications

7.1. Introduction

In terms of stationary power applications, Solid Oxide Fuel Cells (SOFCs) have attracted

considerable interest due to their high efficiencies and fuel flexibility. The most widely

used system within the anode of SOFCs has been Ni dispersed into a matrix of

electrolyte material such as CGO or YSZ [14, 191]. While demonstrating for the most

part excellent performance, this cermet anode is not without its problems, in particular a

tendency for coking at elevated temperatures, as well as sulphur poisoning when using

hydrocarbon fuels such as natural gas. Consequently there has been considerable interest

in the development of new anode materials as outlined in many review articles [160,

161]. In particular, mixed metal oxides that are redox stable are attracting growing

interest, driven by the expectation that they will be less susceptible to the coking or

sulphur poisoning problems associated with Ni [35, 38, 45, 46, 50, 61, 146, 148, 192–

200]. In this respect, perovskite systems of the general formula ABO3 have attracted

considerable interest including Ti, Cr/Mn, and Mo based systems [50, 61, 160, 161].

Doped titanate systems, based around SrTiO3, were one of the first perovskite systems to

be investigated for use within SOFC anodes [35]. Undoped SrTiO3 was found to be

difficult to reduce, even at the high operating temperatures of a SOFC. This difficulty in

reducing SrTiO3 led to low electronic conductivity. However, through doping on either

the Sr or Ti sites, samples with conductivities up to 10 S cm-1

were obtained [45, 199],

163

with further improvements possible via pre-reduction at temperatures up to 1400 °C [38].

This reasonable electronic conduction was not reflected in oxide ion conductivity, which

remained low. This was thought to be in part due to the lack of oxide ion vacancies in the

material.

In separate studies, the perovskite series LaMnO3 has been investigated as a potential

material for use within the cathode of SOFCs [34]. When doped with Sr this series has

high levels of electronic conductivity and is stable at operating temperatures present

within functioning fuel cells. The good electronic conductivity has been attributed to the

mixed valency of manganese, present as both Mn3+

and Mn4+

. Manganese was found to

readily reduce and so these materials were examined further for use within SOFC

anodes, however the ease with which manganese is reduced directly correlates to redox

stability [73, 74].

With a view to trying to improve the conducting properties and stability under reducing

conditions, we have investigated a series of mixed Mn/Ti based perovskites. The starting

point of the series examined is LaMnO3, a traditional SOFC cathode material when

doped with Sr. LaMnO3 itself has orthorhombic symmetry below 750K where it

undergoes a phase change to a cubic system [201]. In order to create a potentially redox

stable material capable of being used as a SOFC anode, the substitution of Ti4+

in place

of Mn3+

has been examined, with charge balance achieved through introduction of Mn2+

on the La3+

site, giving the formula La1-xMnxMn1-xTixO3. The introduction of a mixed

manganese-titanium system is an attempt to improve the oxide ion conductivity over

most titanates while retaining the redox stability of SrTiO3, and also providing an n-type

164

electronic contribution under reducing conditions. This Chapter reports results from the

synthesis of samples in the series La1-xMnxMn1-xTixO3, investigating their conductivities

in air and 5%H2/N2 [202].

7.2. Experimental

High purity La2O3, SrCO3, MnO2, TiO2 were used to prepare the samples of La1-

xMnxMn1-xTixO3-δ (x = 0.2 – 0.5).The powders were intimately ground and heated

initially to 1100 °C for 12h. They were then ball-milled (350 rpm for 1 hour, Fritsch

Pulverisette 7 Planetary Mill with 2 milling beads) and reheated to 1300 - 1350 °C for a

further 12h. Powder X-ray diffraction (Bruker D8 diffractometer with Cu Kα1 radiation)

was used to demonstrate phase purity. Unit cell parameter refinement was carried out

using GSAS suite of programs [123, 124]. Samples were also exposed to 5% H2/N2 for

24 hours at 800 °C to determine their stability in reducing conditions.

Pellets for conductivity measurements were prepared as follows: the powders were first

ball-milled (350 rpm for 1 hour with 2 milling beads, Fritsch Pulverisette 7 Planetary

Mill), before pressing as pellets of 13 mm diameter and sintering at 1300 - 1400 °C for

12h. Conductivities were then measured using the 4 probe DC method in both air and 5%

H2/N2.

165



Figure 49, Figure 50 and Figure 51 show the XRD patterns for several materials in the

series La1-xMnxMn1-xTixO3-δ. Following the synthesis conditions outlined above, the

XRD data showed near single phase samples for x ≤ 0.4, with La2Ti2O7 impurities

observed for higher Ti levels. Table 22 displays the unit cell parameters and volumes for

the series. Small secondary peaks become more apparent at higher values of x, possibly

indicating the formation of a more complex double perovskite system with some

LaMnO3 or MnTiO3 impurity phases present.

x 0 [203] 0.2 0.3 0.4

a (Å) 5.582 5.4938 5.4988 5.5095

b (Å) 5.515 5.5327 5.5500 5.5417

c (Å) 7.763 7.8067 7.7826 7.8329

Volume (Å3) 238.98 237.29 237.51 239.15

Table 22. Unit cell parameters and cell volumes for the series La1-xMnxMn1-xTixO3

As it can be seen, the unit cell expands as x increases. This increase is unexpected; a

decrease is anticipated due to the decrease in average cation size on the A-site, with La3+

having a larger cationic radius of 1.16 Å against Mn2+

at 0.96 Å, while the B-site cation

166

size varies much less with Ti4+

at 0.605 Å against Mn3+

at 0.58 Å and Mn4+

at 0.53 Å

[174]. A possible explanation for this is that there is no manganese present on the A-site;

in this scenario, charge balancing may have been achieved through cation vacancies on

the A site, giving composites such as La1-x/3Mn1-xTixO3. This possibility is reinforced by

noting that the literature values for undoped LaMnO3 show a larger unit cell than the

doped samples, indicating possible vacancies within the unit cell of the doped samples.



167





168

Another interesting detail to note is the small peak at a 2 theta value of ~38°. This shows

the presence of a small quantity of LaMnO3 as detailed in both the Crystallography Open

Database and the literature, helping to verify the formation of an alternate perovskite

system [204]. Many of the peaks for LaMnO3 and LaTiO3 overlap slightly with the main

phase which helps to explain the shoulders and secondary peaks [205].

As expected, the peaks for the samples reduced in 5%H2/N2 shift to lower 2θ values,

showing that some degree of expansion is taking place due to the reduction of one or

more of the cation species present. The XRD patterns for the reduced samples also

suggest that the phases are not fully stable under these conditions as extra peaks are seen.

The peak that can be seen at ~35° and the secondary peak at ~40.5° in the reduced

samples indicates the presence of MnO [206]. This becomes more pronounced as more

Mn is present on the A-site, possibly suggesting the breakdown of the crystal structure or

the presence of an amorphous, Mn-rich phase in the as-prepared samples, which

crystallises on reduction.

7.3.2. Conductivity Measurements


Conductivity measurements were performed on samples of La0.6MnTi0.4O3,

La0.7MnTi0.3O3 and La0.8MnTi0.2O3 (x = 0.4, 0.3 and 0.2). Initial measurements were

taken in air, and these results showed semiconductor behaviour with conductivities in the

169

range 1.6 to 12.5 S cm-1

at 850 °C; the conductivities were lowest for samples with

higher Ti content, as shown in Table 23, Figure 52 and Figure 53.

x 0.2 0.3 0.4

Conductivity (S cm-1

) at 800 °C 11.91 7.62 1.44


) at 700 °C 10.42 6.60 1.13

Table 23. Conductivity data in air for the series La1-xMnxMn1-xTixO3

In addition, the activation energies also increased across the series as the Ti content (x)

increased, Ea = 46.5, 49.2, 59.5 kJ mol-1

for x = 0.2, 0.3, 0.4 respectively as calculated

from Figure 53. The decrease in conductivity in air with increasing Ti content is to be

expected due to the disruption of the Mn-O network. These activation energies are higher

than other similar systems: for example,SrFeO3-δ doped with Mo shows values between

19 - 34 kJ mol-1

in air, while for La0.75Sr0.25Cr0.5Mn0.5O3-δ has activation energies of 20.2

kJ mol-1

in air and 54 kJ mol-1

in H2 [49, 176, 207].

170

Value of x

0.15 0.20 0.25 0.30 0.35 0.40 0.45

Lo

g1

0 C

ond

uctivi

ty (

S c

m-1

)

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

Conductivity in Air

Conductivity in 5%H2/N2

Figure 52. Variation in the conductivities of the series La1-xMnxMn1-xTixO3-δ in air and

5%H2/N2 at 850 °C

171

Figure 53. Log10 σ vs 1000/T for the series La1-xMnxMn1-xTixO3-δ in air

7.3.2.2. Reducing Atmosphere

The conductivities were then measured under 5%H2/N2. Samples were heated in this

atmosphere up to 850 °C and held for 1 hour before making measurements on cooling.

The data showed that the electronic conductivities of all three samples dropped by

several orders of magnitude, to be in the range of 0.015 to 0.4 S cm-1

.

172

x 0.2 0.3 0.4


) at 800 °C 0.33 0.20 0.01


) at 700 °C 0.22 0.11 0.01

Table 24. Conductivity data in 5%H2/N2 for the series La1-xMnxMn1-xTixO3 in 5%H2/N2

Figure 54. Graph of Log10 σ vs 1000/T of series La1-xMnxMn1-xTixO3-δ in 5%H2/N2

As before, the lowest conductivities were observed for the samples with the higher Ti

content. The range of difference between the samples is again apparent, with over an

173

order of magnitude difference between the x = 0.2 and 0.3 and the x = 0.4 samples at

higher temperatures. The activation energies displayed the same trend once more, rising

as x increases; with Ea = 188.0, 154.3, 114.8 kJ mol-1

for x = 0.4, 0.3, 0.2 respectively at

800 °C, considerably higher than alternative materials [49]. Thus in addition to the lower

conductivities, the activation energies are considerably higher than in air, especially for

the high Ti content samples. The samples were analysed by XRD after the measurement,

which confirmed that partial decomposition had occurred. In addition, the observed

conductivities are too low for SOFC anode applications regardless of material redox

stabilities.

7.4. Conclusions

In this work, a new series of perovskite systems has been synthesised, La1-xMnxMn1-

xTixO3-δ (x ≤ 0.4), stable under SOFC cathode operating conditions and displaying partial

decomposition in SOFC anode operating conditions. While reasonable conductivities

were observed in air, the conductivities decreased substantially in reducing

environments, indicating that these systems are not suitable for anode applications.

Work within this Chapter has been published in the International Journal of Low-

Carbon Technologies, Volume 7, Issue 1, Pages 60-62, 2011.

174

Chapter 8. Conclusions and Further Work

Current research suggests that perovskite materials have the potential to display suitable

properties and characteristics as electrode materials for use within SOFCs. The doping of

several systems proved moderately successful, with improvements over the original,

undoped materials being recorded.

Oxyanion doping has proved a particularly viable strategy towards improving the

performance of transition metal containing perovskite materials. In the present case of

doping into SrFeO3-δ, this can partially be attributed to the stabilisation of the cubic

phase in reducing environments, leading to a large increase in conductivity. Further work

on this particular series could include more complete testing within a SOFC, further

compatibility studies with existing electrolyte materials or doping with alternative

oxyanions to study the effect on material characteristics in more detail. This oxyanion

doping strategy has also proved successful in material systems other than perovskites,

such as Ruddlesden Popper phases [64, 65, 67, 68, 171, 173, 208]. There is therefore

wide scope for further work in this field, with a large range of materials potentially

suitable for oxyanion doping.

For the Sr0.8Ti0.6Nb0.4O3-δ materials, vanadium doping was shown to have a positive

effect on conductivity. The best performance was in the material Sr0.8Ti0.6Nb0.2V0.2O3-δ,

where the electronic conductivity in 5%H2/N2 was shown to be fairly stable across a

range of temperatures from 800 to 300 °C, in the region of 2 - 6 S cm-1

, after reduction at

930 °C. The undoped sample had previously shown good stability in reducing

175

conditions, albeit with a resistance to reoxidisation caused by slow oxygen transport

[146]. The V doped samples, in contrast, displayed far faster reoxidisation, even with as

little as 5% vanadium on the B-site, suggesting higher oxide ion conductivity. Thermal

expansion coefficients were shown to be comparable with existing electrolyte materials

such as YSZ, however further work on chemical compatibility with these existing SOFC

materials is needed. Additional work investigating V doping of other Sr1-x/2Ti1-xNbxO3-δ

systems may also give further insight into some of the unexpected cell parameter

changes of Sr0.8Ti0.6Nb0.4-xVxO3-δ.

The La1-xMnxMn1-xTixO3-δ series of materials proved less successful. It proved difficult

to obtain phase pure samples in many cases, with small secondary peaks visible at higher

x values. In addition, after reduction at 800 °C in 5%H2/N2 it can be seen that the

structure is degraded, indicating instability under these conditions. Nevertheless,

reasonable conductivities were obtained in air, with the highest values recorded for the x

= 0.2 sample. Some questions remain over whether there is indeed the incorporation of

Mn onto the A-site, with the possibility of the formation of A-site deficient structures

instead. This would require further verification by other techniques such as combined X-

ray and neutron diffraction studies in order to accurately determine site occupancies.

The extrusion work also proved problematic, with issues surrounding repeatability and

cell performance. While the work carried out appears to indicate the successful formation

of tubes, these suffered from irregularities in wall thickness. In addition, many variations

around material layering and cell preparation failed to produce a successful cell. In order

for additional work to be viable, the extrusion die may need redesigning to allow greater

176

rigidity and so more regular tube sizes, and layering techniques would require further

investigation to provide more repeatable results.

Moving forward, work into perovskite-based SOFCs could be driven forward through

further fabrication of cells in order to form stacks, allowing increased testing profiles to

be carried out on completed cells to better characterise the materials and properties. A

wider selection of fuels, including the ability to introduce potential contaminants and

poisons into the fuel mix, would also allow much greater in-situ analysis of materials

under various performance conditions. The ability to introduce known contaminants into

fuel streams, or even to run directly off the national gas grid, has the potential to prove

the suitability of novel materials and demonstrate in-situ performance.

If in-situ analysis of novel materials was to progress on a larger scale, the ability to

fabricate planar cells would simplify the process and, if a uniform cell size was

established, cell sealing could be made more straightforward which would allow faster

changeover of cells analysed. It would also facilitate more straightforward stack building

if gasket materials and cell interconnects could be established, allowing far greater range

and duration of testing protocols. This work could compliment work on m-SOFCs which

could provide faster verification of suitability of materials due to their better thermal

characteristics.

Integration of SOFCs into buildings as part of CHP systems would allow in-situ testing

of systems and principals for widespread use of fuel cell systems as part of a

microgeneration energy solution. Development of protocols for start-up, shut down and

177

load cycling, amongst other things, could help to build up the requirements for future

progress in the area. In addition, data could be transmitted and processed in real time,

allowing remote monitoring of the system and easing the expansion of such a program

by facilitating centralised system control and development.

Many of these goals could be achieved through the creation of a spin-out company,

supported directly by universities, allowing for transfer of knowledge and access to

facilities while enabling the pursuit of more business- and production-oriented

development. This would allow greater autonomy and the potential to build business

arrangements with larger companies better positioned in the market to roll out future

technologies, possibly to a degree above and beyond that which a university could

support directly. In addition, by focussing less on the development of novel materials,

cells and systems, and more on marketisation, it would have the potential to increase the

speed at which products become commercially viable.

178

Chapter 9. Publications

T. W. Pike, P. R. Slater, International Journal of Low Carbon Technologies, Volume 7,

Issue 1, Pages 60-62 (2011).

J. Porras-Vazquez, T. Pike, C. A. Hancock, J.F. Marco, F. J. Berry, P. R. Slater, Journal

of Materials Chemistry A, Volume 3, Issue 38, Pages 11834 - 11841 (2013).

Chapter 10. References

1. A. B. Lovins, Reinventing Fire: Bold Business Solutions for the New Energy Era,

Chelsea Green Publishing, 2011.

2. F. K. Boese, a. Merkel, D. Stahl, H. Stehle, Solar Energy 26, 1–7 (1981).

3. World Wind Association Half-year Report (2013).

4. J. Bennetsen, Construction VI (2012).

5. Geothermal Energy Association Report, 2013.

6. J. Levene, B. Kroposki, G. Sverdrup, 2006.

7. J. R. Varcoe, R. C. T. Slade, Fuel Cells 5, 187–200 (2005).

8. C. Zuo, S. Zha, M. Liu, M. Hatano, M. Uchiyama, Advanced Materials 18, 3318–3320

(2006).

9. V. M. Goldschmidt, Skrifter Norske Videnskaps-Akademi i Oslo, I. Mat-.-Nat K1

(1926).

10. C. Li, K. C. K. Soh, P. Wu, Journal of Alloys and Compounds 372, 40–48 (2004).

11. C. Greaves, A. J. Jacobson, B. C. Tofield, B. E. F. Fender, Acta Crystallographica

Section B 31, 641–646 (1975).

179

12. B. C. Tofield, C. Greaves, B. E. F. Fender, Materials Research Bulletin 10, 737–745

(1975).

13. H. S. Spacil, Electrical device including nickel-containing stabilized zirconia

electrode, United States, 1970.

14. D. Dees, T. Claar, T. Easler, Journal of The Electrochemical Society 134, 2141–2146

(1987).

15. S. Taylor, Investigation Into The Improved Dispersion of Nickel Oxide Over the

Surface of SOFC Tubular Anodes, University of Birmingham (2010).

16. Y. Matsuzaki, I. Yasuda, Solid State Ionics 132, 261–269 (2000).

17. Z. Cheng, M. Liu, Solid State Ionics 178, 925–935 (2007).

18. N. Akhtar, S. P. Decent, D. Loghin, K. Kendall, Journal of Power Sources 193, 39–

48 (2009).

19. A. Dhir, K. Kendall, Journal of Power Sources 181, 297–303 (2008).

20. A. Dhir, Improved Microtubular Solid Oxide Fuel Cells, University of Birmingham

(2008).

21. H. He, J. M. Hill, Applied Catalysis A: General 317, 284–292 (2007).

22. H. Kan, S. H. Hyun, Y.-G. Shul, H. Lee, Catalysis Communications 11, 180–183

(2009).

23. C. M. Grgicak, R. G. Green, J. B. Giorgi, Journal of Power Sources 179, 317–328

(2008).

24. A. Lashtabeg, S. J. Skinner, Journal of Materials Chemistry 16, 3161–3170 (2006).

25. Z. Lü, L. Pei, T. He, X. Huang, Z. Liu, Y. Ji, Journal of Alloys and Compounds 334,

299–303 (2002).

26. H. Kim, C. Lu, W. L. Worrell, J. M. Vohs, R. J. Gorte, Journal of The

Electrochemical Society 149, A247 (2002).

27. A. Hornés, D. Gamarra, G. Munuera, J. C. Conesa, A. Martínez-Arias, Journal of

Power Sources 169, 9–16 (2007).

28. A. Sin, E. Kopnin, Y. Dubitsky, A. Zaopo, A. S. Aricò, D. La Rosa, L. R. Gullo, V.

Antonucci, Journal of Power Sources 164, 300–305 (2007).

180

29. J. C. Ruiz-Morales, J. Canales-Vazquez, D. Marrero-Lopez, J. Pena-Martinez, A.

Tarancon, J. T. S. Irvine, P. Nunez, Journal of Materials Chemistry 18, 5072–5077

(2008).

30. B. C. H. Steele, Solid State Ionics 129, 95–110 (2000).

31. B. C. H. Steele, Journal of Power Sources 49, 1–14 (1994).

32. H. Yokokawa, H. Tu, B. Iwanschitz, A. Mai, Journal of Power Sources 182, 400–

412 (2008).

33. B. Iwanschitz, L. Holzer, A. Mai, M. Schütze, Solid State Ionics 211, 69–73 (2012).

34. C. S. Tedmon, H. S. Spacil, S. P. Mitoff, Journal of The Electrochemical Society

116, 1170 (1969).

35. P. R. Slater, D. P. Fagg, J. T. S. Irvine, Journal of Materials Chemistry 7, 2495–2498

(1997).

36. D. Fagg, S. Fray, J. Irvine, Solid State Ionics 72, 235–239 (1994).

37. S. Hui, A. Petric, Journal of The Electrochemical Society 149, J1–J10 (2002).

38. S. Hui, A. Petric, Journal of the European Ceramic Society 22, 1673–1681 (2002).

39. D. Neagu, J. J. T. S. Irvine, Chemistry of Materials 23, 1607–1617 (2011).

40. J. Perez-Flores, Chemistry of Materials 25, 2484–2494 (2013).

41. X. Li, H. Zhao, X. Zhou, N. Xu, Z. Xie, N. Chen, International Journal of Hydrogen

Energy 35, 7913–7918 (2010).

42. E. Y. Konysheva, X. Xu, J. T. S. Irvine, Advanced Materials 24, 528–32 (2012).

43. C. Navas, H. zur Loye, Solid State Ionics 93, 171–176 (1996).

44. M. Mori, Z. Wang, T. Itoh, S. Yabui, K. Murai, T. Moriga, Journal of Fuel Cell

Science and Technology 8, 51014 (2011).

45. P. R. Slater, J. T. S. Irvine, Solid State Ionics 120, 125–134 (1999).

46. P. R. Slater, J. T. S. Irvine, Solid State Ionics 124, 61–72 (1999).

47. A. Kaiser, J. L. Bradley, P. R. Slater, J. T. S. Irvine, Solid State Ionics 135, 519–524

(2000).

48. L. Adijanto, R. Küngas, J. Park, J. M. Vohs, R. J. Gorte, International Journal of

Hydrogen Energy 36, 15722–15730 (2011).

181

49. S. Tao, J. T. S. Irvine, Journal of The Electrochemical Society 151, A252–A259

(2004).

50. S. Tao, J. T. S. Irvine, Nature Materials 2, 320–323 (2003).

51. J. Liu, B. D. Madsen, Z. Ji, S. A. Barnett, Electrochemical and Solid-State Letters 5,

A122–A124 (2002).

52. X. J. Chen, Q. L. Liu, S. H. Chan, N. P. Brandon, K. A. Khor, Journal of The

Electrochemical Society 154, B1206 (2007).

53. J. E. Sunstrom, S. M. Kauzlarich, P. Klavins, Chemistry of Materials 4, 346–353

(1992).

54. O. O. A. Marina, N. N. L. N. Canfield, J. W. J. Stevenson, Solid State Ionics 149,

21–28 (2002).

55. D. N. Miller, J. T. S. Irvine, Journal of Power Sources 196, 7323–7327 (2011).

56. V. A. Kolotygin, E. V. Tsipis, A. I. Ivanov, Y. A. Fedotov, I. N. Burmistrov, D. A.

Agarkov, V. V. Sinitsyn, S. I. Bredikhin, V. V. Kharton, Journal of Solid State

Electrochemistry 16, 2335–2348 (2012).

57. S. Hui, A. Petric, Solid State Ionics 143, 275–283 (2001).

58. A. Aguadero, C. de la Calle, J. A. Alonso, D. Pérez-Coll, M. J. Escudero, L. Daza,

Journal of Power Sources 192, 78–83 (2009).

59. L. Zhang, Q. Zhou, Q. He, T. He, Journal of Power Sources 195, 6356–6366 (2010).

60. Y.-H. Huang, R. I. Dass, Z.-L. Xing, J. B. Goodenough, Science 312, 254–257

(2006).

61. Y.-H. Huang, G. Liang, . Croft, . Lehtim ki, M. Karppinen, J. B. Goodenough,

Chemistry of Materials 21, 2319–2326 (2009).

62. Y. Matsuda, M. Karppinen, Y. Yamazaki, H. Yamauchi, Journal of Solid State

Chemistry 182, 1713–1716 (2009).

63. J. F. Shin, L. Hussey, A. Orera, P. R. Slater, Chemical Communications 46, 4613–5

(2010).

64. J. F. Shin, K. Joubel, D. C. Apperley, P. R. Slater, Dalton transactions 41, 261–6

(2012).

65. J. F. Shin, A. Orera, D. C. Apperley, P. R. Slater, Journal of Materials Chemistry 21,

874 (2011).

182

66. C. A. Hancock, R. C. T. Slade, J. R. Varcoe, P. R. Slater, Journal of Solid State

Chemistry 184, 2972–2977 (2011).

67. J. M. Porras-Vazquez, P. R. Slater, Fuel Cells 12, 1056–1063 (2012).

68. J. M. Porras-Vazquez, T. F. Kemp, J. V. Hanna, P. R. Slater, Journal of Materials

Chemistry 22, 8287 (2012).

69. Q. Liu, D. E. Bugaris, G. Xiao, M. Chmara, S. Ma, H.-C. zur Loye, M. D. Amiridis,

F. Chen, Journal of Power Sources 196, 9148–9153 (2011).

70. M. Retuerto, M. Li, Y. Go, A. Ignatov, Inorganic Chemistry , 12273–12280 (2012).

71. R. Martínez-Coronado, A. Aguadero, J. A. Alonso, M. T. Fernández-Díaz, Materials

Research Bulletin 47, 2148–2153 (2012).

72. R. Martínez-Coronado, J. A. Alonso, A. Aguadero, M. T. Fernández-Díaz, Journal of

Power Sources 208, 153–158 (2012).

73. J. Kuo, H. Anderson, D. Sparlin, Journal of Solid State Chemistry 83, 52–60 (1989).

74. N. Q. Minh, Journal of the American Ceramic Society 76, 563–588 (1993).

75. A. Hammouche, E. Siebert, A. Hammou, Materials Research Bulletin 24, 367–380

(1989).

76. S. P. Jiang, Journal of Materials Science 43, 6799–6833 (2008).

77. P. Plonczak, M. Gazda, B. Kusz, P. Jasinski, Journal of Power Sources 181, 1–7

(2008).

78. H. Ullmann, N. Trofimenko, F. Tietz, D. Stover, Solid State Ionics 138, 79–90

(2000).

79. D. M. Bastidas, S. Tao, J. T. S. Irvine, Journal of Materials Chemistry 16, 1603–

1605 (2006).

80. H. Ullmann, N. Naoumidis, D. Stöver, Journal of the European Ceramic Society 19,

791–796 (1999).

81. J. olenda, K. Świerczek, W. Zając, Journal of Power Sources 173, 657–670

(2007).

82. R. Martínez-Coronado, A. Aguadero, C. de la Calle, M. T. Fernández, J. A. Alonso,


83. S. Zha, Z. Cheng, M. Liu, ECS Transactions 1, 293–302 (2006).

183

84. P. Holtappels, F. W. Poulsen, M. Mogensen, Solid State Ionics 135, 675–679 (2000).

85. N. Dai, J. Feng, Z. Wang, T. Jiang, W. Sun, J. Qiao, K. Sun, Journal of Materials

Chemistry A 1, 14147–14153 (2013).

86. a Chang, S. Skinner, J. Kilner, Solid State Ionics 177, 2009–2011 (2006).

87. G. Kim, S. Wang, a. J. Jacobson, L. Reimus, P. Brodersen, C. a. Mims, Journal of

Materials Chemistry 17, 2500 (2007).

88. Y. Li, J. Cheng, J. Song, J. A. J. A. Alonso, M. T. Ferna, J. B. Goodenough, M. T.

Fernández-Díaz, Chemistry of Materials 24, 4114–4122 (2012).

89. G. Amow, I. J. Davidson, S. J. Skinner, Solid State Ionics 177, 1205–1210 (2006).

90. P. Popper, S. N. Ruddlesden, Transactions of the British Ceramic Society 62, 443–

449 (1963).

91. B. Raveau, M. Hervieu, D. Pelloquin, C. Michel, R. Retoux, Zeitschrift für

anorganische und allgemeine Chemie 631, 1831–1839 (2005).

92. R. Chiba, F. Yoshimura, Y. Sakurai, Solid State Ionics 152-153, 575–582 (2002).

93. H. Falcón, A. E. Goeta, G. Punte, R. E. Carbonio, Journal of Solid State Chemistry

133, 379–385 (1997).

94. E. Konysheva, J. T. S. Irvine, Journal of Materials Chemistry 18, 5147 (2008).

95. Z. Shao, S. M. Haile, Nature 431, 170–3 (2004).

96. Z. Z. Shao, W. W. Yang, Y. Cong, H. Dong, J. Tong, G. Xiong, Journal of

Membrane Science 172, 177–188 (2000).

97. S. McIntosh, J. F. Vente, W. G. Haije, D. H. A. Blank, H. J. M. Bouwmeester,

Chemistry of Materials 18, 2187–2193 (2006).

98. M. L. MeCartney, Journal of the American Ceramic Society 70, 54–58 (1987).

99. S. P. S. Badwal, J. Drennan, Journal of Materials Science 22, 3231–3239 (1987).

100. T. Ishihara, H. Matsuda, Y. Takita, Journal of the American Chemical Society 116,

3801–3803 (1994).

101. P. R. Slater, J. T. S. Irvine, T. Ishihara, Y. Takita, Solid State Ionics 107, 319–323

(1998).

102. T. Ishihara, H. Furutani, T. Yamada, Y. Takita, Ionics 3, 209–213 (1997).

184

103. X. Zhang, S. Ohara, R. Maric, H. Okawa, Solid State Ionics 133, 153–160 (2000).

104. R. Cook, R. MacDuff, A. Sammells, Journal of the Electrochemical Society 137,

3309–3310 (1990).

105. X. Tan, W. Yin, B. Meng, X. Meng, N. Yang, Z. Ma, Science in China Series B:

Chemistry 51, 808–812 (2008).

106. T. Jardiel, B. Levenfeld, R. Jiménez, A. Várez, Ceramics International 35, 2329–

2335 (2009).

107. Y. Du, N. M. Sammes, Journal of Power Sources 136, 66–71 (2004).

108. S. P. S. Badwal, F. T. Ciacchi, S. Rajendran, J. Drennan, Solid State Ionics 109,

167–186 (1998).

109. T. Suzuki, M. H. Zahir, T. Yamaguchi, Y. Fujishiro, M. Awano, N. Sammes,


110. N. Droushiotis, M. H. D. Othman, U. Doraswami, Z. Wu, G. Kelsall, K. Li,

Electrochemistry Communications 11, 1799–1802 (2009).

111. X. . Chen, K. . Khor, S. . Chan, L. . Yu, Materials Science and Engineering: A 335,

246–252 (2002).

112. Q. Zhu, B. Fan, Solid State Ionics 176, 889–894 (2005).

113. E. Kendrick, A. Orera, P. R. Slater, Journal of Materials Chemistry 19, 7955–7958

(2009).

114. J. E. H. Sansom, D. Richings, P. R. Slater, Solid State Ionics 139, 205–210 (2001).

115. J. R. Tolchard, M. S. Islam, P. R. Slater, Journal of Materials Chemistry 13, 1956–

1961 (2003).

116. D. Marrero-López, M. C. Martín-Sedeño, J. Peña-Martínez, J. C. Ruiz-Morales, P.

Núñez, M. A. G. Aranda, J. R. Ramos-Barrado, Journal of Power Sources 195, 2496–

2506 (2010).

117. P. Lacorre, F. Goutenoire, O. Bohnke, Nature 404, 9–11 (2000).

118. S. Georges, O. Bohnké, F. Goutenoire, Y. Laligant, J. Fouletier, P. Lacorre, Solid

State Ionics 177, 1715–1720 (2006).

119. M. P. Pechini, US Patent 3.330.697, US Patent, 1967.

120. C. J. Brinker, G. W. Scherer, Sol-Gel Science: The Physics and Chemistry of Sol-

Gel Processing, Academic Press, Inc., 1990, p. 908.

185

121. G. Frenzer, W. F. Maier, Annual Review of Materials Research 36, 281–331 (2006).

122. H. M. Rietveld, Journal of Applied Crystallography 2, 65–71 (1969).

123. A. C. Larson, R. B. Von Dreele, Los Alamos National Laboratory Report LAUR 86

(2000).

124. B. Toby, Journal of Applied Crystallography 34, 210–213 (2001).

125. P. Zoller, John Wiley & Sons, Inc., 2002.

126. C. Hancock, Anion Substitution in Perovskite Related Materials for Fuel Cell

Applications, University of Birmingham (2013).

127. D. Johnson, ZView: a Software Program for IES Analysis, Version 2.8, Scribner

Associates, Inc. Southern Pines, NC, 2008.

128. J. Kiibarsepp, H. Reshetnyak, H. Annuka, J. Kübarsepp, International Journal of

Refractory Metals and Hard Metals 12, 341–348 (1994).

129. S. Lee, J.-M. Kim, H. S. Hong, S.-K. Woo, Journal of Alloys and Compounds 467,

614–621 (2009).

130. H. S. Hong, U.-S. Chae, S.-T. Choo, Journal of Alloys and Compounds 449, 331–

334 (2008).

131. J. J. Benbow, J. Bridgwater, Paste Flow and Extrusion, Clarendon Press, 1993.

132. N. Bailly, S. Georges, E. Djurado, Solid State Ionics 222-223, 1–7 (2012).

133. M. J. Jørgensen, M. Mogensen, Journal of The Electrochemical Society 148, A433–

A442 (2001).

134. A. Martínez-Amesti, A. Larrañaga, L. M. Rodríguez-Martínez, A. T. Aguayo, J. L.

Pizarro, M. L. Nó, A. Laresgoiti, M. I. Arriortua, Journal of Power Sources 185, 401–

410 (2008).

135. P. Blennow, K. K. Hansen, L. R. Wallenberg, M. Mogensen, Solid State Ionics 180,

63–70 (2009).

136. D. Stolten, B. Emonts, Fuel Cell Science and Engineering, Stolten, Emonts, Editors

Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2012;

http://doi.wiley.com/10.1002/9783527650248.

137. M. Suzuki, H. Sasaki, S. Otoshi, A. Kajimura, M. Ippommatsu, Solid State Ionics

62, 125–130 (1993).

186

138. W. Zhou, Z. L. Wang, Scanning microscopy for nanotechnology: techniques and

applications, Springer, 2006.

139. K. Kendall, M. Palin, Journal of Power Sources 71, 268–270 (1998).

140. N. M. Sammes, Y. Du, R. Bove, Journal of Power Sources 145, 428–434 (2005).

141. M. D. Gross, J. M. Vohs, R. J. Gorte, Electrochimica Acta 52, 1951–1957 (2007).

142. Y. Du, N. M. Sammes, G. A. Tompsett, Journal of the European Ceramic Society

20, 959–965 (2000).

143. T. Suzuki, T. Yamaguchi, Y. Fujishiro, M. Awano, Journal of Power Sources 160,

73–77 (2006).

144. K. S. Howe, E. R. Clark, J. Bowen, K. Kendall, International Journal of Hydrogen

Energy 38, 1731–1736 (2013).

145. C. Mallon, K. Kendall, Journal of Power Sources 145, 154–160 (2005).

146. J. T. S. Irvine, P. R. Slater, P. A. Wright, Ionics 2, 213–216 (1996).

147. Y.-H. Huang, R. I. Dass, J. C. Denyszyn, J. B. Goodenough, Journal of The

Electrochemical Society 153, A1266–A1272 (2006).

148. J. C. Ruiz-Morales, J. Canales-Vázquez, C. Savaniu, D. Marrero-López, W. Zhou,

J. T. S. Irvine, Nature 439, 568–71 (2006).

149. J. Yoo, A. Verma, S. Wang, A. J. Jacobson, Journal of the Electrochemical Society

152, A497 (2005).

150. Y. Niu, W. Zhou, J. Sunarso, L. Ge, Z. Zhu, Z. Shao, Journal of Materials

Chemistry 20, 9619 (2010).

151. A. Evdou, V. Zaspalis, L. Nalbandian, Fuel 89, 1265–1273 (2010).

152. M. Hervieu, C. Michel, D. Pelloquin, a. Maignan, B. Raveau, Journal of Solid State

Chemistry 149, 226–235 (2000).

153. S. Ayyappan, V. Menivannan, G. N. Subbanna, C. N. R. Rao, Solid State

Communications 87, 551–552 (1993).

154. P. R. R. Slater, C. Greaves, Physica C: Superconductivity 223, 37–45 (1994).

155. J. P. Hodges, S. Short, J. D. Jorgensen, X. Xiong, B. Dabrowski, S. M. Mini, C. W.

Kimball, Journal of Solid State Chemistry 151, 190–209 (2000).

187

156. V. V. Vashuk, L. V. Kokhanovskii, I. I. Yushkevich, Inorganic Materials 36, 79–83

(2000).

157. M. V Patrakeev, I. A. Leonidov, V. L. Kozhevnikov, V. V Kharton, Solid State

Sciences 6, 907–913 (2004).

158. V. Damljanović, C. Ulrich, C. Bernhard, B. Keimer, P. Mandal, A. Krimmel, A.

Loidl, Physical Review B 73, 172502 (2006).

159. M. Schmidt, S. J. Campbell, Journal of Solid State Chemistry 156, 292–304 (2001).

160. A. Orera, P. R. Slater, Chemistry of Materials 22, 675–690 (2010).

161. A. J. Jacobson, Chemistry of Materials 22, 660–674 (2010).

162. D. Kus c er, D. Hanzel, Journal of the American Ceramic Society 54, 1148–1154

(2001).

163. C. Greaves, P. Slater, Journal of Materials Chemistry 800, 17–21 (1991).

164. C. Greaves, P. R. Slater, Physica C: Superconductivity 175, 172–178 (1991).

165. Y. Miyazaki, H. Yamane, N. Ohnishi, T. Kajitani, K. Hiraga, Y. Morii, S.

Funahashi, T. Hirai, Physica C: Superconductivity 198, 7–13 (1992).

166. A. Maignan, M. Hervieu, C. Michel, B. Raveau, Physica C: Superconductivity 208,

116–120 (1993).

167. D. Pelloquin, M. Hervieu, C. Michel, N. Nguyen, B. Raveau, Journal of Solid State

Chemistry 134, 395–408 (1997).

168. C. C. Appel, N. Bonanos, Journal of the European Ceramic Society 19, 847–851

(1999).

169. M. Aoki, Y.-M. Chiang, I. Kosacki, L. J.-R. Lee, H. Tuller, Y. Liu, Journal of the

American Ceramic Society 79, 1169–1180 (1996).

170. C. A. Hancock, P. R. Slater, Dalton transactions 40, 5599–603 (2011).

171. J. M. Porras-Vazquez, T. Pike, C. A. Hancock, J. F. Marco, F. J. Berry, P. R. Slater,

Journal of Materials Chemistry A 3, 11834–11841 (2013).

172. Multistat Software, Version 1.5, Scribner Associates, Inc. Southern Pines, NC, .

173. J. M. Porras-Vazquez, E. R. Losilla, P. J. Keenan, C. a Hancock, T. F. Kemp, J. V

Hanna, P. R. Slater, Dalton transactions 42, 5397–405 (2013).

188

174. R. D. Shannon, Acta Crystallographica Section A: Crystal Physics 32, 751–767

(1976).

175. M. V. Patrakeev, V. V. Kharton, Y. A. Bakhteeva, A. L. Shaula, I. a. Leonidov, V.

L. Kozhevnikov, E. N. Naumovich, A. a. Yaremchenko, F. M. B. Marques, Solid State

Sciences 8, 476–487 (2006).

176. G. Xiao, Q. Liu, S. Wang, V. G. Komvokis, M. D. Amiridis, A. Heyden, S. Ma, F.

Chen, Journal of Power Sources 202, 63–69 (2012).

177. J. Feng, G. Yang, N. Dai, Z. Wang, W. Sun, D. Rooney, J. Qiao, K. Sun, J. Mater.

Chem. A 2, 17628–17634 (2014).

178. W. Kobsiriphat, B. D. Madsen, Y. Wang, M. Shah, L. D. Marks, S. A. Barnett,

Journal of The Electrochemical Society 157 , B279–B284 (2010).

179. V. V. Kharton, E. V. Tsipis, A. a. Yaremchenko, I. P. Marozau, A. P. Viskup, J. R.

Frade, E. N. Naumovich, Materials Science and Engineering: B 134, 80–88 (2006).

180. V. V Kharton, A. A. Yaremchenko, A. L. Shaula, A. P. Viskup, F. M. B. Marques,

J. R. Frade, E. N. Naumovich, J. R. Casanova, I. P. Marozau, Defect and Diffusion

Forum 226-228, 141–160 (2004).

181. V. V. Kharton, A. L. Shaula, F. M. M. Snijkers, J. F. C. Cooymans, J. J. Luyten, a.

a. A. Yaremchenko, A. a. Valente, E. V. Tsipis, J. R. Frade, F. M. B. Marques, J. Rocha,

Journal of Membrane Science 252, 215–225 (2005).

182. D. Neagu, J. T. S. Irvine, Chemistry of Materials 22, 5042–5053 (2010).

183. V. Kharton, A. Viskup, A. Kovalevsky, Solid State Ionics 133, 57–65 (2000).

184. D. J. Cumming, J. A. Kilner, S. Skinner, Journal of Materials Chemistry 21, 5021

(2011).

185. P. Blennow, A. Hagen, K. Hansen, L. Wallenberg, M. Mogensen, Solid State Ionics

179, 2047–2058 (2008).

186. J. C. Pérez-Flores, D. Pérez-Coll, S. García-Martín, C. Ritter, G. C. Mather, J.

Canales-Vázquez, M. Gálvez-Sánchez, F. García-Alvarado, U. Amador, Chemistry of

Materials 25, 2484–2494 (2013).

187. J. Karczewski, B. Riegel, M. Gazda, P. Jasinski, B. Kusz, Journal of

Electroceramics 24, 326–330 (2009).

188. W. Ito, T. Nagai, T. Sakon, Solid State Ionics 178, 809–816 (2007).

189. P. Vernoux, M. Guillodo, J. Fouletier, A. Hammou, Solid State Ionics 135, 425–431

(2000).

189

190. N. Danilovic, J.-L. Luo, K. T. Chuang, A. R. Sanger, Journal of Power Sources

194, 252–262 (2009).

191. O. Yamamoto, Y. Takeda, R. Kanno, M. Noda, Solid State Ionics 22, 241–246

(1987).

192. K. R. Kendall, C. Navas, J. K. Thomas, H.-C. zur Loye, Solid State Ionics 82, 215–

223 (1995).

193. J. Canales-Vázquez, S. W. Tao, J. T. S. Irvine, Solid State Ionics 159, 159–165

(2003).

194. J. Karczewski, B. Riegel, S. Molin, a. Winiarski, M. Gazda, P. Jasinski, L.

Murawski, B. Kusz, Journal of Alloys and Compounds 473, 496–499 (2009).

195. E. Kendrick, M. Islam, P. Slater, Solid State Ionics 176, 2975–2978 (2005).

196. F. Bidrawn, G. Kim, N. Aramrueang, J. M. Vohs, R. J. Gorte, Journal of Power

Sources 195, 720–728 (2010).

197. a Ovalle, J. Ruizmorales, J. Canalesvazquez, D. Marrerolopez, J. Irvine, Solid State

Ionics 177, 1997–2003 (2006).

198. M. J. Escudero, J. T. S. Irvine, L. Daza, Journal of Power Sources 192, 43–50

(2009).

199. Q. X. Fu, F. Tietz, D. Stöver, Journal of The Electrochemical Society 153, D74–

D83 (2006).

200. J. C. Ruiz-Morales, J. Canales-Vázquez, C. Savaniu, D. Marrero-López, P. Núñez,

W. Zhou, J. T. S. Irvine, Physical Chemistry Chemical Physics 9, 1821–1830 (2007).

201. J. Rodriguez-Carvajal, M. Hennion, Physical Review B 57, 3189–3192 (1998).

202. T. W. Pike, P. R. Slater, International Journal of Low-Carbon Technologies 7, 60–

62 (2011).

203. S. Husain, R. J. Choudhary, R. Kumar, 58, 1045–1049 (2002).

204. Y. Guo, X. Zhang, R. Wappling, R. Wäppling, Journal of Alloys and Compounds

306, 133–140 (2000).

205. M. Kestigian, R. Ward, Journal of the American Chemical Society 76, 6027 (1954).

206. K. Fujino, S. Sasaki, Y. Takéuchi, R. Sadanaga, Acta Crystallographica Section B

37, 513–518 (1981).

190

207. A. A. Markov, I. A. Leonidov, M. V. Patrakeev, V. L. Kozhevnikov, O. A.

Savinskaya, U. V. Ancharova, A. P. Nemudry, Solid State Ionics 179, 1050–1053

(2008).

208. J. F. Shin, P. R. Slater, Journal of Power Sources 196, 8539–8543 (2011).

Documents

Development and processing of solid oxide fuel cell materialsetheses.bham.ac.uk/5861/1/Pike15PhD.pdf · processing of novel materials for use within solid oxide fuel cells. A range